1The standardised reporting of banks’ lending, key sectoral measures already analysed across banks, and GHG emissions mapped to sectoral lending is all the necessary information for the main analysis of interest: construction of two climate-relevant indices, Emissions Exposure Index (EE_i) and Emissions Funding Index (EF_i) for each bank i.

4.1 Definition of the indices

4.1.1 Emissions Exposure Index (EE_i)

2This is a measure of a bank’s portfolio climate risk exposure. The index answers the question “How much is each bank exposed to climate risk through its sectoral loan composition, given the GHG emissions represented by its loan portfolio?”

3Recall that GL_ij denotes the gross loans by each bank i to each economic sector j and S_j is the share of emissions attributable to each sector. Also, we defined W_ij as the share of each bank’s loan portfolio to each sector, such that where the index k runs over the set of all sectors and where ΣW_ij = 1 for each bank.

4The Emissions Exposure Index (EE_i) is the sum of the relative exposures of each bank i across all economic sectors j weighted by the relative share of emissions attributable to each sector S_j, that is:

EEi = Σ_jSjWij (3)

5For example, assume Bank A has its entire lending concentrated on agriculture (W_A,AGR = 1) and Bank B directs its lending entirely to financial services (W_B,FIN = 1); since the share of emissions in the agriculture sector is higher than in financial services, i.e. S_AGR > S_FIN, then EE_A > EE_B in this case.

6Thus, EE_i captures the GHG emissions intensity of the loan portfolio of each bank according to the emission shares represented by its sectoral loan composition.

4.1.2 Emissions Funding Index (EF_i)

7This is a measure of the climate risk funded by the bank’s lending portfolio. The index answers the question “How much of the climate risk attributed to GHG emissions does each bank fund through its lending relative to other banks?”

8Again, GL_ij denotes the gross loans by each bank i to each economic sector j and S_j is the share of emissions attributable to each sector. In addition to W_ij, we also defined M_ij as the market share of each bank in each sector, such that where the index k runs over the set of all banks and where ΣM_ij = 1 for each sector.

9The Emissions Funding Index (EF_i) is the sum of the market shares of each bank i across all economic sectors j weighted by the relative share of emissions attributable to each sector S_j, that is:

EFi = Σ_jSjMij (4)

10ΣEF_i = 1 across all banks.

11For example, assume there are only two Banks, A and B, that lend entirely to the transport sector with market shares of 40% and 60% respectively (M_A,TRAN = 0.4, M_B,TRAN = 0.6); since the share of transport sector emissions S_TRAN is the same for both banks, the difference in market shares means EF_A < EF_B as Bank B funds a larger proportion of emissions.

12Thus, EF_i captures the relative funding of climate risk and is a measure of climate risk importance of each bank, given the emissions funded by its lending relative to other banks.

13As noted by Monasterolo et al. (2017) in their novel development of climate-relevant indices for the euro area, these two indices are complementary and help capture two very important aspects of climate risk for Kenyan banks: the climate risk exposure represented by banks’ portfolios and their relative funding of the climate risk through their lending. No such related indices have been constructed for banks in any other emerging economies in Africa. Further, no available research on climate-related disclosures for banks has mapped GHG emissions to lending sectors using the hybrid approach applied in this study.

4.2 Results and interpretation

14Figure 1 below reports the values of EE_i and EF_i calculated for each bank across all sectors.

Figure 1: Results of EE_i and EF_i analysis

Emissions Exposure Index, EE_i

15From the main results in Figure 1 above, all banks except one have EE_i values around a narrow range of 0.06–0.09. Given the index definition, this means that the banks have a fairly similar exposure to GHG emissions based on their sectoral lending composition. CFC Stanbic has a relatively higher EE_i of 0.14. As the sample is representative of the industry, this implies a fairly similar concentration of climate risk exposure across Kenyan banks given their sectoral loan composition. The relatively higher exposure by CFC Stanbic can be explained by the high concentration of its loan portfolio in the high-emission agriculture and real estate sectors as seen in Table 7 (collectively accounting for 51% of the bank’s gross loans).

16The key finding from the Emissions Exposure Index (EE_i) analysis can be summed up as follows: Kenyan banks, with the exception of one outlier, have fairly similar exposure to climate risk through their loan portfolios, given the GHG emissions represented by their sectoral lending.

Emissions Funding Index, EF_i

17From the main results in Figure 1, there is a differentiation in the values of EF_i across the banks which is fairly proportional to their market share of gross loans. Given the index definition, this means that the banks have differentiated funding of climate risk through their lending that is fairly proportional to their market shares. CFC Stanbic has a high emissions funding relative to its size (EF_i of 16% similar to the biggest bank KCB) and this can be explained by its relatively high climate risk exposure EE_i seen earlier. The differentiation of emissions funding that mimics market shares is not unexpected given that banks have a fairly similar concentration of exposure as concluded from the EE_i analysis. Thus, bigger (smaller) banks have higher (lower) funding of climate risk.

18The key finding from the Emissions Funding Index (EF_i) analysis can be summed up as follows: Kenyan banks, with the exception of one outlier, have differentiated funding of climate risk attributed to their loan portfolios that is fairly proportional to their market shares of gross loans, and this is expected given the similarity in their similar exposure profiles.

Rescaling emissions funding to 100%

19From the main results in Figure 1, values of EF_i add up to 0.89, i.e. the funding of sectoral emissions through the banks’ loan portfolios add up to 89% of emissions attributed to the sectors collectively. This is because there are emissions within the economy that are not directly linked to any of the bank lending sectors. These are mostly the “other economic activities” identified in the GDP reporting that are not directly linked to any of the distinct lending sectors. They include economic activities in the following GDP categories: public administration; professional, administration and social services; education; health; and other services.

20The calculation of EE_i and EF_i is repeated this time rescaling the emissions sectoral contribution such that the lending sectors account for 100% of emissions funding within the banking industry. Figure 2 uses the rescaled emissions, i.e. ΣEF_i = 1 to report the values of EE_i and EF_i. As the relative weighting of emissions across sectors is unchanged, the interpretation and conclusion is the same as in Figure 1: Kenyan banks, with the exception of one outlier, have differentiated funding of climate risk attributed to their loan portfolios that is fairly proportional to their market shares of gross loans.

Figure 2: Results of EE_i and EF_i analysis (rescaled emissions=100%)

21From Figure 2, the non-sampled banks account for 13% of emissions funding implying that the sample of banks representing 82% of industry gross loans account for 87% of emissions holdings.

4.3 Forward-looking scenario analysis

22The values of EE_i and EF_i are known for each bank, together with the respective interpretation of the two indices for the banking industry. To enhance the analysis, I am interested in a forward-looking view of the expected values of the indices in 2030. This helps enable a futuristic view of the climate risk exposure across the banks. Two scenarios are of interest:

• a business-as-usual (BAU) scenario that uses the emissions forecasts for 2030;

• a transition scenario of abating emissions by 30% relative to 2030 BAU.

23For both scenarios, I repeat the analysis of EE_i and EF_i using the BAU emissions forecasts and the transition scenario 30% emissions reduction target as shown in column 1 and column 2 respectively in Table 12 below. For both cases, the futuristic view is determined with regard to the emissions (and not the sectoral loan composition), and thus we use the current sectoral loan composition for each bank but with the 2030 emissions forecasts for each scenario.

4.3.1 2030 business-as-usual scenario

24Figure 3 below reports the values of EE_i and EF_i using the 2030 BAU emissions forecasts.

Figure 3: Results of EE_i and EF_i analysis – 2030 BAU scenario

25From the results in Figure 3 above using 2030 BAU emissions forecasts, most EE_i values are still within a narrow range, 0.07–0.12 with CFC Stanbic still as the outlier. In addition, there is one more result of interest: there is an emissions exposure spike for I&M Bank and NIC Bank (which is now merged with CBA Bank to form NCBA). This EE_i spike is explained by high lending exposure to the manufacturing, energy, and water sector (24% sector share by each of the two banks). The energy sector is forecast to have the highest increase in emissions by 2030, with a 12-fold increase from 4 MtCO2e in 2015 to 47 MtCO2e in 2030 (see Table 12), accounting for 63% of the increase in emissions over the 15-year period. This significant increase in the risk exposure of the two banks with a high concentration in lending to the manufacturing, energy, and water sector is associated with a small increase in the funding risk EF_i. Overall, EF_i results are still consistent with the earlier finding that larger (smaller) banks have higher (lower) funding of climate risk.

4.3.2. 2030 transition scenario

26The transition scenario is expected to have an effect on the values of the indices in 2030 given the increase in expected emissions but also the fact that different sectors have different abatement potential and thus this affects S_j, the relative share of emissions attributable to each sector in 2030. Like with the BAU scenario, I use the current sectoral loan composition for each bank but with the 2030 emissions forecasts with the 30% abatement target scenario. Figure 4 below reports the values of EE_i and EF_i using the 2030 transition scenario

Figure 4: Results of EE_i and EF_i analysis – 2030 transition scenario

27From the results in Figure 4 above using 2030 transition emissions forecasts, most EE_i values are still within a range, 0.05–0.11 with CFC Stanbic still as the outlier. In addition, just like in the 2030 BAU scenario, there is an emissions exposure spike for I&M Bank and NIC Bank (which is now merged with CBA Bank to form NCBA). This spike is still explained by the high lending exposure to the manufacturing, energy, and water sector, which has expected a significant increase in emission levels. While there is a 21% drop in the sector’s forecast emissions from 47 MtCO2e to 36 MtCO2e under the 2030 reduction targets, the latter still represents a significant increase in emissions from 4 MtCO2e in 2015. For contrast, the LULUCF sector has a forecast 93% emissions reduction from 26 MtCO2e in 2015 to 2 MtCO2e under the 2030 transition scenario. Thus, these two banks that have a high concentration in lending to the manufacturing, energy, and water sector are still expected to have a significant increase in risk exposure with an associated small increase in the funding risk. Overall, like with the 2030 BAU scenario, the EF_i results are still consistent with the earlier finding that larger (smaller) banks have higher (lower) funding of climate-related risk.