Emission reduction targets and outcomes of the Clean Development Mechanism (2005–2020)

The Clean Development Mechanism (CDM) allows developing countries to earn carbon credit units by reducing greenhouse gas emissions. Here we assess the emission reduction outcomes of the CDM between 2005 and 2020. The analysis covers 3,311 CDM projects hosted by 79 countries and over 10,000 Monitoring Reports. We identify which host countries and project types departed from original forecasts more. Overall, the total amount of actual emission reductions was 16% below the targets envisaged by project proponents. Emission reduction projects consistently under-performed over the year, but performance varied between and within regions. Industrial HFCs and N 2 O projects exceeded their targets, whereas landfill gas and methane avoidance projects under-performed by larger mar-gins. Economic gains were unevenly distributed. Estimated revenues relative to GDP were higher for larger emerging economies, and disproportionately smaller for the deprived members of the Global South. Four host countries (China, India, South Korea and Brazil) not only dominated the market, but also gained an advantage from the higher carbon prices before 2012. Least Developed Countries had their carbon credits issued in more recent years when prices were much lower. The results show an imbalance in economic outcomes and raise questions about the effectiveness and equity of this Kyoto mechanism. Weak targets under Paris Agreement could intensify these challenges.


Introduction
Article 6 of the Paris Agreement provides a basis for establishing new mechanisms for mitigating greenhouse gas (GHG) emissions after 2020. The new mechanisms are likely to involve the use of internationally transferred mitigation outcomes to achieve nationally determined contributions (NDCs). These mechanisms are expected to replace the Clean Development Mechanism (CDM), a multi-billion euro carbon finance mechanism established under the Kyoto Protocol [1]. The CDM enables Annex I Parties (industrialized countries) to produce and acquire carbon offset units through investments in GHG mitigation projects in non-Annex I Parties, predominantly in the Global South. These offset units, known as Certified Emission Reductions (CER), represent a reduction, avoidance or sequestration of one metric ton of CO 2 that would otherwise be emitted into the atmosphere. Annex I Parties can use CERs to meet their emission reduction targets under the Kyoto Protocol [2,3]. The CDM issued the first CERs in October 2005. Since then, it has mobilized more than US$162 billion of financing to non-Annex I Parties. As of September 2021. US$162 billion is the total capital investment by CDM projects, including Programmes of Activities (PoAs), that have been issued CERs. This is a conservative estimate, because not all projects reported their financial details. The US$162 billion is estimated from a fraction of projects with CERs issued (N = 2,786) that collectively accounts for 69.5% of estimated (ex ante) annual emissions [5].
This article assesses the emission reductions outcomes of 3,311 CDM projects that have been issued CERs. The main objective is to measure project performance by comparing actual emission reductions (ex post) with estimated emission reductions (ex ante). There were signs of under-performance as early as 2006 [4], but evidence is conclusive only when most projects are given time to operate to their proposed timeframe. Most projects have a crediting period of either 7 or 10 years, and most of them were registered on or before 2011. We identify the project types and host countries that did not meet the emission reduction targets stated in CDM project proposals, known as Product Design Documents (PDDs). To determine how many CERs to issue, the CDM Executive Board (EB) requires the project proponent to submit a Monitoring Report (MR), which specifies the amount of estimated and actual emission reductions in a given monitoring period. The data we used are derived from these MRs published between 2005 and 2021, and an official CDM project database.
The CDM was also designed to channel carbon finance to the Global South. The size of direct economic contributions to non-Annex I Parties depends on the quantity and price of CER units. Our analysis compares the price expectations of project proponents and market CER prices to understand the discrepancies in economic contributions. We show the extent to which direct economic contributions relative to GDP, are unevenly distributed between higher-and lower-income developing countries. We explain this in terms of the timing of CER issuance and price volatility. Our study covers the active lifetime of the CDM to provide conclusive evidence and insights into the ways forward. The findings raise questions about the effectiveness of the CDM in meeting its targets and delivering equitable outcomes. CDM-style Paris mechanisms may result in similar patterns of uneven distribution of benefits, if their scope and designs largely replicate those of Kyoto mechanisms. These results are particularly important for re-thinking the limits of these mechanisms in delivering development benefits and how they should be designed in order to benefit the economically deprived members of the Global South in more direct and effective ways.

Methods
The timeframe of our analysis, which runs from 2005 to 2020, is important for understanding our findings. The first commitment period of the Kyoto Protocol commenced in 2005. The first few years after 2005 witnessed a market boom with higher CER prices. Sharp changes occurred towards 2012, when the first commitment period completed. The end of 2020 is another watershed. The second commitment period concluded on 31 December 2020. The European Union no longer accepts the use of international units, including CERs, for compliance under the EU ETS after 2020. While the CDM continues to operate, no new projects are expected to seek registration.
Emission reduction data were collected from MRs. Other project information was gathered from the official database of CDM projects (cdm.unfccc.int/Projects/projsearch.html), including project type, host Party (country), duration of monitoring period, date of CER issuance, and start date of validation. The database was accessed in March 2021 and updated in September 2021 [5].

CDM projects included in the analysis
The full CDM database has 8,206 registered CDM projects. Our project-level analysis excluded 352 Programmes of Activities (PoAs), because new projects can be added to the programme without undergoing the complete CDM project cycle, and therefore it is hard to attribute emission reduction estimates to their component Project Activities. PoAs account for 2.2% of all CERs issued [5]. A total of 3,312 projects (out of 7,854), formally known as CDM Project Activities, have been issued CERs. The first set of carbon units was issued in October 2005, and the latest date of issuance was 30 August 2021. We excluded one project (CDM reference #1) and the first 19 Monitoring Reports (MRs) of another project (CDM reference #3), because their MRs did not clearly specify the amount of estimated emission reductions for the relevant monitoring period and their PDDs did not include a breakdown of yearly estimates. The project activities of these two projects (first 19 MRs of CDM Project #3) generated a total of 66,958,005 CER units. Our analysis on emission reductions is based on the remaining 3,311 CDM projects, which were hosted by 79 non-Annex I Parties.

Estimating emission reductions and measuring performance
The amount of emission reduction is stated in the MRs of these 3,311 projects. Each project has at least one MR published, and the average is four MRs. There are over 10,000 MRs. Typically, each Monitoring Report has only one issuance record, but those that transcend the first and second Kyoto commitment periods have two records. We reviewed each MR and recorded the amount of actual emission reductions reported by the project proponent, which was verified by an independent auditor and accepted by the CDM EB before CERs were issued. The majority of MRs were revised after reviews, and some of them were withdrawn. We adopted the revised and final versions of MRs and excluded withdrawn MRs from the analysis. Each MR indicates the amount of emissions actually reduced during a given monitoring period, against a ex ante estimate for the same period. We identified 13 MRs that indicate their actual emission reductions exceeding the amount of CERs issued. These 13 MRs reported more actual emission reductions than CERs issued. A common explanation is that the CDM EB requires the amount of CERs issued be capped at the level of average annual emissions estimated in the registered PDD. In analyzing emission reductions, we adopted the value of emissions actually reduced, as reported in the MR. As a result, the aggregate values do not perfectly match the total amount of CERs issued. Duration of monitoring period ranges from 1 days to 4,107 days and averages 419 days (N = 10,999). The start date of the earliest monitoring period was 1 April 2000, and the latest end date was 31 December 2020.
Ex ante estimates are also extracted from MRs. However, some of the pre-2010 MRs do not indicate the amount of estimated emission reductions for the corresponding monitoring period. To document the original expectations of project proponents, we retrieved these emission reduction targets from the first PDD that was approved for registration, rather than the post-registration, revised PDD. Moreover, we did not take the self-reported estimates at face value, but cross-checked the calculation presented in the MR against the approved PDD to ensure accuracy and consistency. We believe that there are typos, miscalculations, and tendencies for strategically choosing variables (e.g., number of actual operating days) to present a lower value of estimated emission reductions in MRs. For some projects, the amount of estimated emission reductions varies every year. We therefore adopted the estimates specific to the monitoring period concerned, rather than the yearly average of the entire crediting period, where applicable. For this reason, we did not use the estimated emission reduction data from the official CDM Database. This database also does not provide data on actual emission reduction, which are gathered from individual MRs.
Performance rate is calculated to indicate the extent in which a project met its pre-determined emission reduction targets within the crediting period(s) in which CERs were issued. It is the ratio of actual emission reductions to estimated emission reductions. Weighted performance rates are used to describe projects (N = 3,311). Each project resulted in a different amount of emission reductions, and therefore its impact on the overall performance rate of a host country or project type is weighted by the total amount of emission reductions generated from this project during the crediting period(s). Average performance rates (not weighted) are used to describe CER issuance records (N = 10,999).

Comparing carbon prices
We compared the price expectations of project proponents with the market CER prices. A sub-sample of the 3,311 projects was used. An expected price is the level of market CER price predicted by the project proponent. It is the offer price at which they expected to sell their CER units. Price expectations are a key driver of CDM investment decision-making. The expected level of CER price is often stated in the PDD or its attachments for demonstrating the project's expected higher financial performance over an economic baseline (i.e., return on investment without CER incomes) in order to justify that the investment would create 'additional' emission reductions, especially for renewable energy projects [6,7]. We identified expected prices from 2,113 CDM projects. Where multiple expected prices were presented in a PDD, the lowest one was adopted as a conservative estimate (i.e., the closest to market prices, in most cases).
Weighted reference prices are calculated on the basis of market CER prices. It is used in this study to approximate the market value of a CER unit at the time when it was issued, and to compare with the target (i.e., expected price). Our calculation was based on an average value of the daily CER closing prices in the 30 trading days prior to the date of CER issuance. We adopted the prices of monthly CER futures contracts traded in the Intercontinental Exchange, because there was very limited spot CER trading from the end of 2012. Monthly CER futures prices are the closest alternative to spot prices. Spot price is the current price at which a CER unit can be traded for immediate delivery. The price set in a monthly futures contract is usually very close to its spot price, due to the short timeframe. Yet, we were unable to retrieve this price information for CER units issued before March 2008, which account for 5.7% of total CER units issued. We used annual average secondary CER prices obtained from the World Bank's reports (2007,2009) to substitute for these missing prices.
Weighting is based on emission reductions. A CDM project might have its CERs issued at different times during or after the crediting period. Accordingly, there were multiple reference prices for the different monitoring periods of the same project. The reference prices of each project were weighted by the amount of actual emission reductions the project generated in the corresponding monitoring period. As a result, these prices reflect the timing of CER issuance. Because CER prices were on a downward trend after 2011, the more the CER units issued in later years, the lower the weighted prices.
These prices are plotted in a graph and ordered by the start date of project validation, which marks the time when the project proposal was submitted for third-party validation [8]. Same as the 'Start of Project's Public Comment Period' specified in [8]. The official database of CDM projects includes complete information on the validation start date of each project, which is therefore reliable and consistent for the purpose of research [5].

Estimating economic contributions
The volume of CER units generated relative to the GDP of a given host country is used by the UNEP DTU Partnership as "an immediate expression of the importance of CDM to the economy" [9]. The ratio of these two variables provides an indication of the direct economic benefits of producing and exporting these units, given the economic scale of the host country. We used the amount of actual emission reductions for estimating economic contributions, but also included revenue estimates to account for price variations.
We estimated revenues (ex post) from sale of CER units for each of the 3,311 projects, which are the product of actual emission reductions in a given monitoring period and the reference CER price (without weighting). Because we used secondary market prices as reference prices, the value of CER units generated from 'bi/multilateral' CDM projects might have been underestimated. Bi/multilateral projects have a formal arrangement at the time of project development with an Annex I Party that intends to buy and use the credits. They operate under a forward purchase agreement, which typically specifies a mutually agreed price level or a price range at which the CER units generated from the project would be sold to a buyer from an Annex I Party. Purchase agreements with set prices could protect project proponents and host countries from volatile prices. On the other hand, 'unilateral' CDM projects do not have a letter of approval from an Annex I Party at the time of registration and do not have a prior contractual arrangement with an international buyer [2].
We do not have access to all forward transaction data and therefore cannot include agreement prices in our analysis. Despite this limitation, our revenue estimates are useful for comparing economic contributions between countries. Many emission reduction purchase agreements are guaranteed only up to 2012 [4], but over 45% of CER units were issued after 2012. Agreement on a transaction price was typically reached long before the project started and therefore likely to be higher than market prices, which began to fall sharply from 2011. The large differences between the two prices since 2011 challenge the assumption that all purchase agreements would be exercised at the indicative prices and volumes [10]. CER units might eventually be sold outside the range of these agreement prices-likely at a lower level, given the availability of significantly cheaper CER units in the market. Renegotiations and cancellations did occur. Official data suggest that 19% of buyers have withdrawn from the bi/multilateral projects to which they were committed [9]. Moreover, 69.6% of all CDM projects were bi/multilaterally established (Table 1), but China alone accounts for most of these projects and has an exceptionally high percentage of bi/multilateral projects. Excluding China, less than half of the CDM projects were bi/multilaterally established. Although market prices do not perfectly reflect actual transaction prices, the changes in the latter track market trends. This is important for the purpose of this analysis, which aims to highlight the time-sensitive economic contributions of the CDM in a period of over 15 years.
The revenues analysis adopted the amount of CERs issued, rather than actual emission reduction, because only CERs are traded in the market. All revenue estimates were discounted by 2% to account for the transfer of proceeds from CER sales to the Adaptation Fund established under Kyoto Protocol. To address the different price levels at different times of CER issuance, the aggregate revenue estimates were converted to 2010 prices using the International Monetary Fund (IMF) Euro Area GDP deflators for the time (year) of CER issuance.
For estimating economic contributions to individual host countries, the IMF GDP estimates were adopted as denominators and expressed in Euro (converted from USD using year-end exchange rates). The ratio between revenues (current prices) and GDP was calculated for each country and each year between 2005 and 2021, using the nominal GDP estimate for that year. The values were then aggregated for each country. Project type, geographical region, Commitment Period, host country and whether it is a 'Least Developed Country' (LDC) were specified in the CDM database.

Summary
The key measures and indicators used in the analysis are listed below: Actual emission reductions: as stated in CDM Monitoring Reports. They indicate the amount of net GHG emissions reduced, avoided or sequestered in the combined monitoring periods of a project and determine the amount of CERs issued. Combined monitoring periods are the fraction of crediting period(s) in which CER units were issued.
Estimated emission reductions: as stated in CDM Monitoring Reports and PDDs. These estimates are a fraction of the total estimated emission reductions of a CDM project and represent what was expected to achieve in the combined monitoring periods. The remaining emission reductions to be generated in the rest of the crediting period(s) (i.e., CERs yet to be issued) was excluded.
Performance rate: ratio of actual emission reductions to estimated emission reductions.
Expected CER price: as stated in PDDs. Representing the offer price expectation of the CDM project proponent and the expected revenues per CER credit sold. Non-Euro currencies were converted to Euro, at the exchange rate at the start date of validation.
Weighted reference price: average value of the daily prices of monthly CER futures contracts over the 30 trading days prior to the date of CER issuance. The average values calculated for each project were weighted by the amount of actual emission reductions the project generated in its corresponding monitoring period. Estimated revenues: product of the CER units issued in a given monitoring period and the reference price at the date of CER issuance. Discounted by 2% and converted to 2010 prices.
Economic contributions: ratio of 1) total amount of actual emission reductions, and 2) estimated revenues, to the GDP of a given host country.

Emission reductions
The 3,311 CDM projects reduced 2,043 million tonnes of emissions, which is 16.4% below the aggregate PDD forecasts (2,444 million). Fig 1 shows  , most of their projects achieved 80% of their target or more. African LDCs generally performed better than the rest of the world, but the rest of Africa was not. It should be noted that these discrepancies can be attributed to inflated emission baselines or poor project performance, or both. Project types explain some of these results.  These projects involve destruction of HFCs and N 2 O gases from industrial processes, which are of high global warming potential and generate a disproportionately large amount of emission reductions. There are only 79 HFCs and N 2 O projects, but they produce 10.6 million tonnes of CO 2 e each on average and account for 41% of total actual emission reductions from all projects. South Korea and China generated 85.0% and 44.6% of their emission reductions from these projects respectively, much higher than India (24.9%). These three countries were responsible for over 80% of all HFCs and N 2 O-based emission reductions. Excluding these 79 projects from the sample (N = 3,311) reduced the overall performance rate from 84% to 74%. Despite their effectiveness in meeting emission reduction targets, HFCs projects are highly controversial and have been called into question for creating perverse incentives for manufacturing more of these gases to increase baseline emissions, by which to maximize CER revenues. It is likely that part of the claimed or reported emission reductions are not real and additional, raising question about the environmental integrity of these projects [11,12]. These projects are no longer accepted for CDM registration since 2007.
The majority of CDM projects (1,950) produce hydro and wind power. Few of these projects achieved their emission reduction target. Their MRs have suggested several reasons. Emission factors were declining in some countries, meaning that the burning of fossil fuels becomes less polluting over time and therefore renewable energy displaced fewer emissions than proposed. Weather is a more common explanation. Seasonal variations are foreseeable, but the availability of water (rains) and wind is more often overestimated than less. Another reason, which is not always explicitly stated in MRs but widely discussed elsewhere, is a mismatch between policy intervention and actual need. China, for example, produced 67.4% of all emission reductions based on wind, hydro, and solar power. The Chinese government has considerably increased the incentives for installing renewable energy capacities, mobilizing a massive amount of private and public capital into these industries. However, the installed capacity booms did not meet with actual energy demand and improvements in the capacity of supporting infrastructure, resulting in excess supply and eventually curtailment of wind, hydro, solar power across the country [13][14][15]. The reduced operation of power generating facilities inevitably left more emissions unabated.
Landfill gas and methane avoidance projects recorded the lowest performance rates among project types with more than 100 projects issued CERs. Both of them involve the handling and disposal of waste, but landfills reduced more emissions. Landfill gas projects involve the capture and degradation of the methane and CO 2 produced during landfill waste decomposition by flaring or use for power generation. Mexico achieved only 67% of its ex ante emission reductions. Fifty-one Mexican projects, out of 72, involved either landfill gases or methane avoidance. Project performance was remarkably low. One of these projects (CDM reference #1371) commenced in 2008, but achieved only 3.5% of the pre-project estimate during a fiveyear monitoring period from 2015, citing non-operation in 1,531 days (out of 1,824 days). On average, these 51 projects departed from their targets by over 50%.
Their MRs attribute poor performance to factors such as low efficiency of the gas capturing systems, gas release resulting from exposure of historical waste during infrastructure installation, and lower baseline emissions due to a change in local regulations causing decreased waste disposal (CDM reference #1123, #2992, and #4211). Some landfill sites begin as unmanaged dumping grounds which are not run by the government and therefore historical records of landfill activity can be unreliable or inaccessible, making it difficult to estimate the historical waste composition which is a key determinant of landfill gas production [16]. New legislations and policies that promoted waste minimization also resulted in unexpectedly smaller volumes of disposed waste delivered to the landfill sites, directly reducing the baseline GHG emissions that would have been released into the atmosphere (CDM reference #4211). Other reasons listed in MRs (all project types) include poor estimation before project implementation, accidents, maintenance and repairs, natural hazards, and changes in economic conditions. CDM projects consistently under-performed over time. This is clearer for countries and regions with more issuance records available, including China, India, and the rest of Asia and Latin America, which account for 81.1% of all CER issuance records (Fig 3). Similarly, the average performance rates of wind power, hydropower, methane avoidance, biomass energy, and landfill gas projects varied before 2010 but became steady afterwards (Fig 4). These five project types account for 74.3% of all issuance records. The fewer emissions actually reduced directly lead to a smaller amount of carbon finance. This is exacerbated by the shrinkage of the CDM market from 2011.

Carbon prices
Price expectations and market prices moved in the same direction up to a point. Fig 5 displays the movement of expected prices and weighted reference prices between 2003 and 2020, along with the daily closing prices of CER units. The discrepancies between expected and market prices increased since the middle of 2008. CDM projects that began validation at a later time were less likely to reach the forecasted price when their CER units were available to the market. In 2011, it became clear that CER units could only be used to a limited extent in the phase 3 (2013-2020) of the EU ETS. This triggered a collapse of CER prices, as the EU ETS was the largest buyer of CER units. While the CER prices struggled to return to higher levels, a small number of projects proposed around this time continued to show confidence in the price. There were tendencies for overestimating CER prices, or being too optimistic.
The decrease in price disadvantaged countries that issued CERs in more recent years, notably the LDCs. This can be illustrated by mapping estimated CER revenues.

Economic contributions
The direct economic contributions of the CDM are unevenly distributed. Nonetheless, the differences are relatively small, indicated by the low contrasts in filled colours (Fig 6). About 53.5% of CER units were issued on or before 2012. Emerging economies, particularly China, earned most of the CER units, but some LDCs held a fair share of the market. The main driver was the EU ETS, which would accept new project credits/CERs after 2012 only if the project is registered in one of the LDCs. As a consequence, the demand for CER units from non-LDCs diminished, and investment in CDM projects hosted by a LDC increased. The biggest beneficiaries were Asian LDCs. Bhutan, Cambodia, and Lao PDR hosted 19 CDM projects and earned 903, 339, and 286 CER units per million GDP (Euro), respectively. Hydropower contributed to 98.3% of the CER units issued to them.
Total revenues for the first and second Kyoto commitment periods are estimated as €10,480 million and €173 million (2010 prices) respectively. Host countries do not receive a fair share proportionate to their economic scale, because their CER units were issued in different periods of time. The decrease in CER price intensified the uneven distribution of benefits between large emerging markets and small LDCs. Fig 7 shows the amount of estimated CER revenues relative to nominal GDP. The clearer contrasts in colour reflect the effects of price and timing of CER issuance. China stands farther apart from other non-Annex I Parties by generating a disproportionately large return from the CDM. India, South Korea, and Brazil are three other larger national beneficiaries. Revenues was also high for lower-middle income economies (non-LDCs), such as Nicaragua, Vietnam, and Bolivia. In contrast, these estimates were remarkably lower for Asian LDCs, including Bhutan, Cambodia, and Lao PDR, and African LDCs, notably Uganda and Lesotho.
The small pie charts at the bottom of Figs 6 and 7 suggest that China, India, South Korea and Brazil together earned 81.1% of all CER units issued, but accounted for 90.2% of the estimated revenues (or 90.5% for the first Kyoto commitment period). This implies that the average market value of their units is higher than those of the rest of the world. The differences between China and other host countries are likely to be larger than presented in Fig 6, because China has an exceptionally high proportion of bi/multilateral partnerships which could shield revenues from sliding prices-much higher than its 'competitors', namely, India, South Korea, and Brazil (see Table 1 above). Therefore, the economic benefits of the CDM are highly concentrated in a handful of emerging economies, especially China.
The timing of project and CER issuance contribute to an uneven distribution. As indicated in Fig 8, emerging economies had a large proportion of their CER units issued on or before 2012, followed by the rest of Latin America (Fig 8B). Benefits to these countries were comparatively attractive, thanks to the higher prices in the first few years of CDM implementation. LDCs were underrepresented in the market until 2012, when they gained momentum from European and international policy interventions that supported LDC participation (Fig 8C). However, revenues from the post-2012 market were lower, even if the sale prices were protected by purchase agreements, as most of which were established in the later years and reflected the declining market conditions. This remains true, but to a lesser extent, for the rest of Africa and Asia (Fig 8D). The direct economic contributions of the CDM were disproportionately small for most African and Asian host countries other than China, India, South Korea, and Vietnam.

Discussion and conclusions
New carbon crediting mechanisms under the Paris Agreement present an opportunity for redressing some of the previously unforeseen or underestimated problems of Kyoto mechanisms. The CDM is part of the global action to create a cleaner and more equitable future by financing emission reduction activities in developing countries. CDM projects that have been issued CERs are in principle able to produce 4.6 billion CERs to the end of 2020 [17]. This Kyoto mechanism has presented a steep learning curve for its successors [18,19].
A total of 3,311 registered CDM projects have reduced, avoided or sequestered a minimum of 2,043 million tCO 2 e. This is 16% below the original PDD forecasts, or 26% lower if the controversial HFCs and N 2 O gas projects are excluded. Carbon prices moved in the opposite direction as CDM project proponents forecasted. The extended period of low prices could reduce the incentives for operating the project and bring it to a halt. In this situation, the project proponent could manage to support the project using internal resources or through conventional financing [6,20], implying that international finance through the CDM would no longer play a decisive role-even if the project continues to reduce emissions. Furthermore, the economic benefits of the CDM were concentrated in emerging economies. Direct benefits to LDCs were disproportionately smaller, because more of their CER units were subject to the frustrating prices.
The results shed light on the financing outcomes of the CDM. The amount of carbon mitigation finance mobilized did not meet prior expectation. For instance, in October 2009, the World Bank anticipated that the CDM could raise €13.5 billion (US$18 billion) in direct carbon revenues for developing countries by the end the first Kyoto commitment period [21]. Our conversative estimate (€10.5 billion) is 22.2% lower. This is conservative because secondary market prices are typically higher than the direct financial benefit to host countries [20]. According to our data, the pre-2012 weighted average price of CERs was EUR 12.49 (constant 2010 prices). Annual reports of the World Bank suggest that the weighted average price in the primary CER market between 2005 and 2012 was EUR 9.37 (constant 2010 prices) [15, 20, 27,  28]. This raises questions about the effectiveness of the CDM in mobilizing finance, considering that the mechanism itself was a significant global investment and involved massive resources and efforts in negotiation, coordination, and operation.
The financing outcomes do not appear equitable. The World Bank anticipated that China, India, South Korea and Brazil would receive 78.3% of direct carbon revenues by the end of the first commitment period [21], whereas our study suggests 90.5% for the same period. Large emerging economies have considerable advantages over the poorest ones. For example, the Chinese government required local project proponents to identify an Annex I Party partner before seeking national approval and therefore many Chinese CDM projects are bound to purchase agreements. This could protect the country's interests from the price collapse since 2010, and reflects the stronger institutional support provided by the Chinese government for building international partnership for CDM activities [22,23]. LDCs, on the other hand, did not gain any advantage during the market boom before 2012. The post-2012 rules that prioritized LDCs met with a market downturn. Earlier research has raised

PLOS CLIMATE
questions to the ability of the CDM to deliver development benefits [24,25]. We echo these previous studies by suggesting that the design of the CDM, market volatility, and differential capacities for project development result in a 'development deficit' that Paris mechanisms must address.
The unmet expectations are concerning in the wake of the Paris Agreement. Systematic under-performance can be an indication of certain problems and loopholes in the processes of project registration and auditing gone undetected. There have been reports of inflation in baseline emissions and price expectations, and strategic manipulation of project variables in order to demonstrate feasibility and additionality [6,8,[26][27][28]. Under-performance can also arise from poorly designed or managed projects. If these problems and loopholes are actually more significant and prevalent than previously documented, the performance of the mechanism may be even lower than reported here. A strong rulebook under Article 6 and measures for enhancing regulatory effectiveness would be important for ensuring the reliability of project proposals and audit reports, and the performance of projects.
Unlike the Kyoto Protocol, the Paris Agreement requires Parties to formulate their NDCs, which are voluntary emission reduction targets. The bottom-up approach for determining emission reduction targets can lead to a lack of ambition in controlling domestic emissions and consequently weaken the aggregate demand for carbon units [19,29]. Sharp reductions and movements in carbon price as a result of changes in national or international commitments could marginalize economically deprived members of the Global South in the processes of project development, implementation, and transferring carbon units. LDCs arguably encounter more barriers than large emerging economies. As recipients of carbon finance, their economic interests in the international transfer of carbon units risk being compromised by the volatility of the market, which can arise from a unilateral decision to retreat from climate action by dominant GHG emitting countries and a rapid decline in demand from these countries. Ambitions in emission reduction, programme continuity, and additional institutional support to economically deprived Global South countries will be crucial for delivering equitable outcomes under the new crediting mechanisms.