【论文精读】Bridging socioeconomic pathways of CO2 emission and credit risk

1 Introduction

The context of CO2 emission scenario.

Without proactive measures to limit these CO2 and other GHG emissions, one can expect a global warming by 3 or 4oC, and maybe more, by 2100. This current global warming is and will continue to profoundly disrupt environmental, geographical and economic balances, if no mitigation and adaptation measures are taken. The Paris Agreement at the twenty-first session of the Conference of the Parties (COP 21) is an important milestone in international climate policy as it establishes a global mitigation framework towards 2030 and sets the ground for a global warming with stabilization around 1.5oC only. This idealized scenario is based on a carbon neutrality around 2050, with some variations depending on the countries according to their Nationally Determined Contributions (NDC). Actually there are many other scenarios, depending on the ecological transition trajectory that countries, economic actors and populations will follow. In the most recent scientific literature, these scenarios are known as Shared Socioeconomic Pathways (SSPs)

See Figure 1 for the global CO2 emission in different sectors, in the Organization for Economic Co-operation and Development (OECD), according to the scenarios chosen, the data are available on the SSP Public Database https://tntcat.iiasa.ac.at/SspDb.

Climate risks in finance

Climate change generates new sources of risk (so-called climate risks), in particular physical and transition risks as described by the solemn resounding speech by Mark Carney.

In this work, we mainly focus on the transition risks and explore how to link projection scenarios such as those described in the SSPs by Phase 6 of Coupled Model Intercomparison Project (CMIP6) into credit risk projections for firms. More specifically, we provide a quantitative model where the inputs are some desired paths of CO2 emission (et)t_0, the production characteristics of a company, its sensitivity to CO2 emissions, its climate-free credit spread and the outputs are the stochastic evolution of credit spread in an uncertain commercial demand.

We consider the firm who aims to maximize its production profit and at the same time takes into consideration the CO2 reduction plan described by SSPs. Over-emission compared to the target may induce penalty. From the point of view of the firm, the objective is to determine the optimal strategy of its effective emission by solving a penalized optimization problem. The credit quality of the firm can be impacted by such carbon emission transition via its cash flow. In the classic structural credit approach such as Merton or Black-Cox model, a default event occurs when the firm’s value is inferior to its debt level. We describe the firm’s value process as the discounted value of all its future cash flow according to the optimal emission and production profit and study the corresponding default probability.

The optimal production problem in order to maximize expected profit of a company is well studied. We suppose that the firm’s production depends on its energy consumption and in particular on the carbon emission level. We consider in addition the over-emission penalty under certain probabilistic risk measure  where a loss function is concerned.  We then follow the classic structural models in credit risk to compute the climate-related default probability of the firm and show by numerical examples the impact of different relevant parameters and SSPs scenarios on the default probabilities and intensities.

2 Model and results

2.1 Production and carbon emission constraint

a probability space(Ω, A,P)

We consider a firm whose production P=(Pt;t≥0) depends on the energy consumption and in particular on its effective CO2-emission volume γ=(γt;t≥0)and solves the following stochastic differential equation (SDE),

where γt is the instantaneous emission rate at time t, Wt is an F-Brownian motion which represents the uncertainty in demand and supply for the production and σ is a positive constant volatility parameter. μ characterizes the production rate and satisfies the local Lipschitz condition x for an independent positive constant K.

Empirical studies show that overproduction can lead to a decrease of production rate for example due to excess of supply, whereas the effect of emissions on production growth is positive. We thus suppose that μ is decreasing with respect to the production P and increasing with respect to the emission γ.

In response to the greenhouse gas emission reduction target, the firm has an objective emission plan described by e=(et;t≥0), which represents the emission benchmark suggested by a SSP projection or accredited to the firm by the European Commission. This quantity could be deterministic or stochastic meaning respectively the allowances fixed or recommended with certain tolerance by regulation.

In this paper, we consider the effective emission which is in general supposed to be positive. However, in other context such as carbon sequestration, or the emission allowance compensation within the European Emission Trading System (ETS), the SSPs may take negative values (see e.g. Figure 1 Energy sector). In other words, the objective emission et may be negative, meaning that it has to be considered as net emission, once that all processes related to compensation and carbon capture have been taken into account.

Loss function

The firm’s goal is to maximize its production profit and, at the same time, manage the effective emission level by taking into account the advertised constraints.

Given a benchmark emission projection, the firm chooses its optimal effective emission to maximize the expected profit by controlling the related production, cost, and emission constraints. The optimization problem is presented as below. We consider the profit maximization over all future time under the pathwise emission constraint and define the objective optimization function as

where r > 0 is a constant2discount rate. We aim to solve

 

2.2 Optimal emission strategy

Profit maximization in an explicit model

 

 

2.3 Credit risk under emission transition

In this section, we study the credit risk of the firm induced by the transition towards the low carbon emission and production pattern. We use the effective production obtained in the previous optimization problem to deduce the firm’s value process and then compute the default probability in a structural modelling approach.

 

 

 

Structural credit model and default probability

In a structural credit model, the firm is considered to default when its value process gets below a default threshold.

We let the default barrier be described by a deterministic function L(t) which depends on time and represents the minimal level of the firm’s liability payment such as the debt reimbursement together with labor and other functioning costs (as the operational and capital expenditure) at time t. When the firm’s value is higher than the threshold, then it is in a financially sustainable situation and can function normally. In the contrary case, the firm encounters fiscal difficulty and a default event may be triggered.

 

3 Numerical illustrations

3.1 SSPs scenarios and optimal emission

We consider the following CO2 emission scenarios which correspond to different socioeconomic reference pathways provided by CMIP6: SSP1-2.6, SSP2-4.5, SSP3-LowNTCF, SSP4-6.0, and SSP5-3.4-OS. These scenarios are illustrative pathways adopted by the IPCC in the sixth Assessment Report indicating the CO2 concentrations in atmosphere from lowest (SSP1) to highest (SSP5).

In other words, SSP1 and SSP5 describe respectively economic growth pattern via sustainable and fossil-fuel pathways. We choose two sectors: Transportation (Figure 2) and Industrial (Figure 3) sectors for which the year 2015 is our starting point. For each sector, we consider the above five SSPs including two baseline scenarios: SSP1-2.6 which is the most mitigated scenario corresponding approximately to the previous scenario generation Representative Concentration Pathway (RCP) 2.6, and SSP2-4.5 with is a moderate scenario similar to RCP-4.5. We also consider three (Tier 2) supplementary scenarios: SSP3-LowNTCF (Near-Term Climate Forcing) which provides a comparison scenarios with high NTCF emissions (notably SOx and methane), SSP4-6.0 focusing on a socio-economic context of inequality, and SSP5-34-OS (OverShoot) which allows for large overshoot by mid-century followed by substantive policy tools in the latter half of the century.

 

 

 

 

 

3.2 Default probability for different sectors

We now present the emission-related default probability in the explicit case and analyze the impact of carbon emission reduction given different SSPs benchmarks. We still consider the explicit model in Proposition 1 and use a slightly different definition for the value process associated to the optimal

As expected the initial value of the intensity coincides with the prefixed value λ0=3%. When time evolves, the more constrained scenarios are associated to larger default probabilities and higher intensities. The SSP1-2.6 scenario is the most impacted one, which is quite natural given its immediate and hard reduction strategy. The scenario which follows is SSP5-3.4-OS: although this benchmark allows for a large overshoot up to 2060, the relatively strict mitigation during the latter period makes the default probability and intensity increase significantly. Observe that SSP4-6.0 corresponds to a fixed intensity of λ0=3% in the Transportation sector and the same phenomenon appears for the SSP3-LowNTCF scenario in the Industrial sector, as the optimal emission is unconstrained in these two cases. We note that in this study we only investigate the transition risk related to the firm’s mitigation strategy and ignore the possible physical risks under each scenario for example the more frequent damage and natural catastrophes under scenarios with higher temperature increase such as a SSP5 scenario. 

Figure 7 illustrates the default intensity for Industrial sector under the impact of the parameter c representing the dependence of the firm on emission and ω representing the penalty force. We consider the hardest scenario SSP1-2.6 and a moderate one SSP4-6.0. For both scenarios, the increase of one of the two parameters implies a higher default intensity and this phenomenon is particularly accentuated when the firm’s production is highly dependent on the CO2 emission (when c is large). For such a firm, a strong penalty policy together with a hard mitigation scenario such as SSP1-2.6 (left) could have a significant impact on the firm’s default probability.