A climate investment trap in developing economies

13 Finance is vital for the green energy transition, but the access to low cost finance is uneven 14 as the cost of capital differs substantially between regions. This study shows how modelled 15 decarbonisation pathways of developing economies are disproportionately impacted by 16 assumptions around their cost of capital (WACC). For example, representing regionally 17 specific WACC values indicates 35% lower green electricity production in Africa for a cost- 18 optimal 2°C pathway. Moreover, results show that early convergence of WACC values for 19 green and brown technologies in 2050 would allow Africa to reach net-zero emissions 20 approximately 10 years earlier than when convergence is not considered. A “climate 21 investment trap” arises for developing economies when climate-related investments remain 22 chronically insufficient. Elements of sustainable finance frameworks currently present barriers 23 to these finance flows and radical changes are needed so that capital is better allocated to the 24 regions that most need it. 25

chronically insufficient. Elements of sustainable finance frameworks currently present barriers 23 to these finance flows and radical changes are needed so that capital is better allocated to the 24 regions that most need it.   Note: The strength of these links is strictly linked to local conditions implying that the set of self-reinforcing 122 mechanism could be exacerbated (or less relevant) in some economies.

124 125
The cost of capital in our scenarios 126 In this analysis, we introduced region-specific cost of capital values (here-after referred to as

138
We study scenarios that achieve the 2°C target -rather than the 1.5°C target -to examine the WACCs are uniform across regions but differ between green and brown electricity generation,

147
and are set at the mean global values, weighted by GDP, of 5.9% and 5.1% respectively. The 148 'Regional' scenario (REG) employs the regional WACCs, which vary between green and 149 brown technologies, as shown in Figure 2. The WACC is generally higher for green 150 technologies and in developing economies. In the third and fourth scenarios, C2050 and 151 C2100, the WACC values used in the REG scenario converge by 2050 and 2100, respectively.

152
For each region, the values converge to whichever is lower -the global average value for 153 brown technologies or the lower WACC of that region ( Figure 2

Results 164
We examine how modelling regional WACCs and different speeds of WACC convergence 165 impact electricity decarbonisation pathways and investments in developing and developed 166 economies. To highlight the implications for representative countries with high and low risk 167 profiles, we focus on the results for Western Europe and Africa, which face the principal 168 challenges of replacing fossil fuel infrastructure with green power technologies, and of scaling

194
The impact of regional and globally uniform WACC values on CO2 emissions is greatest for 195 regions with regional WACCs that deviate the most from the uniform WACC. Net-zero 196 emissions in Africa is achieved in 2058 in the GBL scenario and in 2065 in the REG scenario, 197 while the difference is negligible in Western Europe. Our estimates show that representing the 198 observed local financing conditions leads to regional higher emissions in Africa (20% in 2050) 199 due to the lower renewable investments deployed.

201
The striking consequence of implementing regional versus globally-uniform WACC values is 202 that decarbonisation pathways in developing economies are highly affected. Under the REG scenario, they register a much lower (globally cost-optimal) level of renewable deployment 204 and a slower rate of emissions reduction than in the GBL scenario. Convergence between green and brown WACC values has a significant impact on the 210 electricity generation mix in Africa, especially when WACC convergence is achieved by 2050 211 (figure 4). In 2050, in the C2050 and C2100 scenarios, green electricity production in Africa is 212 43.1% and 6.5% higher than the REG scenario, respectively. If convergence were brought 213 forward, there would be considerably more green electricity production in the first half of the 214 century. The REG and C2100 scenarios follow similar pathways until 2050, with the C2100 215 pathway leading to slightly more green electricity thereafter, moving toward the levels of green 216 generation seen in the C2050 scenario by 2100. Once again, the relative difference in Western  The timing of the WACC convergence has a large impact on green power investments in Africa 220 (figure 5    Our results suggest that a more rapid WACC convergence will allow developing economies 255 to achieve much higher level of green electricity deployment and faster emissions reduction.

Policy implications 265
The results show that earlier WACC convergence could allow developing regions to reach net-   In the space of a few months, economies across the world have been severely stressed, with 366 developing economies exposed most of all due to their relatively low resilience. This is 367 particularly true for economies whose income is highly dependent on fossil fuels, such as 368 some countries in Africa (e.g. Nigeria). Nonetheless, the pandemic may also present a window 369 of opportunity to reframe the global commons (including public health and the climate system) 370 with respect to international market finance, in which a lower cost of capital for developing 371 economies would allow for low-carbon development at a more internationally equitable cost.

386
The WACC thus provides a simple measure to evaluate perceived risks for a given investment 387 in a specific market. In this study, the WACC is used as a proxy to assess the differences in 388 risk-premiums associated with energy assets across countries and/or regions, as it represents 389 the weighted average of the costs of raising funding (equity and debt) for a specific investment 390 Miller 1958, 1963). The cost of equity depends on the risk that equity investors 391 perceive in the project in a specific market, while the cost of debt reflects the default risk that 392 lenders perceive from the same investment in that market (Kumar 2016

500
In all scenarios, GDP growth is calibrated to match the Shared Socioeconomic Pathway SSP2.

501
In addition, energy demand is driven by the SSP2 population growth and the structure of the

506
The modelling analysis is based on scenarios achieving a 2°C target by the end of the century.
temperature rise can exceed 2°C during the model timeframe but it must return to reach 2°C 509 or lower in 2100. In all the modelled pathways that have the temperature limit applied, the 510 global temperature rise reaches 2.23°C in 2060.

REG
2C in 2100 Regional (Fig 2) and constant over the period GBL 2C in 2100 Uniform and constant over the period after 2020

Annex -Global results 518
Global results for the different scenarios are presented in this section. We focus on power 519 generation and provide the total (including brown and green power) and green electricity 520 production.     2,691  8,440  17,708  34,800  53,889  60,901  63,869  66,007  70,448   Green  WEU  GBL  3,799  6,165  8,529  15,466  20,667  21,588  21,278  22,202  23,068   Green  WEU  REG  3,799  6,125  8,550  15,355  20,404  21,317  21,208  21,486  22,536   Green  WEU  C2050  3,799  6,155  8,665  15,794  20,356  21,061  21,125  21,467  22,627   Green  WEU  C2100  3,799  6,159  8,580  15,400  20,291  21,266  21,214  21,839  22,610   Total  AFR  GBL  2,911  3,266  4,009  7,682  13,830  24,107  36,295  44,240  48,274   Total  AFR  REG  2,911  3,338  6,213  11,251  20,137  30,189  40,023  49, Figure 1 The climate investment trap. Note: The strength of these links is strictly linked to local conditions implying that the set of self-reinforcing mechanism could be exacerbated (or less relevant) in some   Regional investment levels in green electricity for the 2°C scenarios with different convergence years. Absolute values (left) and changes compared to the REG scenario (right) Figure 6 Regional CO2 emissions for the 2°C scenarios with different convergence years