Do Carbon Offsets Offset Carbon?

Are carbon offsets actually offsetting carbon?

Raphael Calel, Jonathan Colmer, Antoine Dechezleprêtre, and Matthieu Glachant examine the misallocation of carbon offsets in their paper "Do Carbon Offsets Offset Carbon?".

They build a database of 1,346 Indian wind farms constructed between 1992 and 2013, of which 476 were registered under the Clean Development Mechanism (CDM), and develop a partial-ranking method to identify projects that would very likely have been built without any subsidy.

Their main conclusions include:

• Roughly 54% of subsidised wind projects are « inframarginal » and look strictly more profitable ex ante than unsubsidised competitors built in the same state and year.
• These projects collectively received around 27 million carbon credits, more than half of all offsets approved for Indian wind power, and account for 4.22 GW of subsidised capacity that would likely have materialised regardless.
• When the authors simulate allocating subsidies at random, the actual CDM allocation directs more credits to inframarginal projects than chance would in 99.6% of simulated draws.
• If extrapolated programme-wide, the CDM may have approved 6.1 billion tonnes of CO2 worth of offsets, with discounted economic losses valued near $1.2T using a $190 social cost of carbon.
• Sensitivity analyses all preserve the central finding, with at least 5 million inframarginal credits surviving even under the most stringent specifications.
• Diplomatic cables and audit records suggest a layered institutional failure: the Indian authority takes developers at their word, third-party verifiers are paid by the projects they assess, and the CDM Executive Board approved over 95% of registration requests.

This article suggests ex ante additionality tests on carbon offsets built on internal rate of return computations cannot be trusted when the underlying assumptions are supplied by the developers themselves.

If the world's largest compliance offset programme misallocates over half of its credits in a sector where additionality is easiest to defend, standardised benchmarks and stronger ex post verification appear essential for emerging crediting standards.

Generalising the 52% rate to other technologies, countries, and to voluntary credits however requires caution, since structural conditions and verification practices vary substantially across programmes.