Short answer
Location based, not the “market-based”.
Long explanation
Leading experts in GHG accounting have rejected the WRI/WBSCD GHG Protocol’s “market-based” method for Scope 2 GHG accounting as being fundamentally flawed.[1] [2] This rejection is because this method, at its core, allows an organization to report Scope 2 emissions based upon a financial transaction that does not alter its physical consumption of energy or the emissions physically associated with its operations or assets. Emissions that are physically associated with its electricity consumption, and therefore properly attributed to the organization, are represented by a location-based average grid emission factor because the electrical energy on a grid is undifferentiated and undifferentiable with respect to its origin.
Further, even under a consequential accounting method, the voluntary purchase of RECs and GOs by companies and consumers have been clearly shown to not cause emission reductions (see Should I use RECs or GOs to calculate my organization’s carbon footprint?), and therefore, these transactions do not result in benefits for the environment, which could be claimed by a consumer.
Note that corporate GHG accounting (attributional) Scope 2 estimates that utilize the “market-based method” also ignore line losses (see Should RECs or GOs be used for any form of GHG emissions accounting?). This mismatch is one more indication that RECs were not designed for and are not appropriate for GHG accounting purposes.
[1] Open letter rejecting the use of contractual emission factors in reporting GHG Protocol Scope 2 emissions (2015). Available here: https://scope2openletter.wordpress.com
[2] Brander, M., Gillenwater, M., and Ascui, F. (2018). Creative accounting: A critical perspective on the market-based method for reporting purchased electricity (scope 2) emissions. Energy Policy.