Structurally, this study follows a similar approach to the CBECI by introducing hypothetical lower and upper bounds that form the range of estimated GHG emissions. This is the basis of the sensitivity analysis, which is summarised in Figure 1. The two limits are outlier scenarios based on the abstract assumption that all the electricity used is generated by a single energy source (coal-only or hydro-only). This format is simply illustrative and provides context to make interpreting the results more intuitive. In reality, the electricity used is generated by a mix of energy sources (the electricity mix). Knowing the composition of this mix is essential to computing the emission intensity associated with the Bitcoin network. In this study, the emission intensity serves as a measurement of GHGs emitted per kilowatt-hour, given as gCO₂e/kWh. Table 3 lists all the energy sources used in this study with their respective emission intensities.

While the emission intensity remains constant for both outlier scenarios, as only one energy source powering the Bitcoin network is assumed, the best-guess estimate accounts for all the energy sources in Table 3. This presupposes an understanding of which energy sources are used and to what extent. Therefore, determining the emission intensity of the entire Bitcoin network requires data on the relevant geographical locations of the mining facilities from which the computing power (hashrate) originates and their electricity mix.

Besides showing a range of estimates given different assumptions on the energy sources used, a solution had to be found to deal with periods for which information was unavailable. As mentioned previously, the geographical location of mining facilities is vital to determining the electricity mix. However, data on the geographical distribution of hashrate only started to be collected in September 2019 and is available only until the most recent update of the mining map. Therefore, a solution had to be found for the period before data became available, and for the months for which data was not yet available. To address this challenge, ann. GHG emission estimates are calculated for three different time intervals – historical, assessed and predicted – as shown in Figure 2.

The issue of unavailable data during certain periods only applies to the best-guess estimate, since the upper- and lower-bound estimates are based on a single energy source (coal-only or hydro-only) and, hence, changes in geographical hashrate distribution do not impact the emission intensity. By assuming a single energy source, the emission intensity stays constant over time and always corresponds to the value in Table 3, regardless of where the hashrate originates.

For the best-guess estimate, depending on the time interval, different approaches had to be developed to estimate GHG emissions. The estimates for the historical and predicted intervals are based on major simplifications, while computing estimates for the assessed interval is more complex:

• The historical interval assumes that Bitcoin mining operations are distributed proportionally across the world, based on each country’s share of total global electricity production.
• The assessed interval allows for much greater granularity as, during this period, data on global hashrate distribution is available monthly.
• The predicted interval assumes that the most recent mining map update is still a valid approximation of the current hashrate distribution.

The assumptions that underpin each interval have significant implications for the estimated emission intensity during that interval. As illustrated in Figure 3, the emission intensity fluctuates only slightly during the historical interval and remains constant during the predicted interval. However, when global or regional changes in Bitcoin mining activity are considered, as they are in the assessed interval, much greater fluctuations in emission intensity become visible. A more detailed explanation of these assumptions, including the calculations for all our estimates, is provided in the following sections.

In the upper-bound scenario, it is assumed that all the electricity consumed by Bitcoin miners is generated by coal. In the lower-bound scenario, the assumption is that all the electricity is generated by hydropower. Therefore, the emission intensities of the upper- and lower-bound estimates always equal the emission intensity of the relevant energy source.

The model also provides estimates of Bitcoin's underlying electricity consumption by source and emissions intensity. Both estimates are derived from geographical data on Bitcoin mining operations and regional data on energy sources used for electricity generation.