Weathering the Storm: An Introduction to Climate Risk with MATLAB

View Example

Melting polar ice caps, a depleting ozone layer and rising oceanic temperatures aren’t the only things affected by climate change – financial institutions could pay the price, too. In this article, we give a brief introduction to what climate risk is, how it can be a component of financial risk, and how this could be applied within the MATLAB programming environment.

The impacts of climate change can be far-reaching and devastating. According to the United Nations, 2019 was the hottest year on record and the global average temperature in 2019 was 1.1 C above the pre-industrial period [1]. Further, total annual greenhouse gas emissions reached its highest ever levels in 2018, with no sign of peaking. In addition, carbon emissions are set to reach 56 gigatonnes of CO2 by 2030 – that’s almost double what they should be [1]. These can cause influence farming yields, cause extreme weather events and result in a critical loss in biodiversity.

However, one question might be that as far-reaching and destructive as climate change is, how does that affect the financial sector? Simply put, the effects of climate change affect the stability of financial institutions by disrupting and causing uncertainty to regular activities, which in turn affect their investors and the economy. Changes in policy could result in depreciation of assets and, therefore, models need to be adapted to account for or combat the depreciation.

In South Africa, the Carbon Tax Act No 15 of 2019, came into effect from 1 June 2019 [2]. This law uses the polluter-pays-principle: where large emitters of greenhouse gases are penalised more for production of these gases. This forces these companies to take these costs into account for future planning and investment decisions; with the goal that it would incentivise the use of clean technologies.  These added costs reduce the potential operating profit, and therefore could negatively affect share prices and possible dividends in the short term and sustainability of the business in the long term  As a portfolio manager, your approach to asset allocation may be affected. You could choose to invest with companies who have made the move to become “greener” and thus incur less penalties, or to avoid companies whose stocks have been affected by the implementation of climate policy. With 479 “green bonds” issues in 2019 – a 125% increase from 2018 – the trend of “green investing” is surely here to stay [3].

Changing governmental policy and extreme weather events increase the uncertainty in the financial stability of businesses which is a concern investors and credit providers alike. One way to quantify uncertainty has traditionally always fallen under risk modelling: hence the designation of climate risk. We have seen growing interest from our customers around climate risk.

The modelling of meteorological events is being explored as an indicative measure to diagnose the level of climate risk associated to either investment or credit decisions. When it comes to modelling risk –climate or financial – MATLAB gives you the flexibility to model any type of risk – whether it be credit, counterparty, market or operational.

MATLAB can be leveraged as a powerful simulation engine to perform Monte Carlo simulations, for example. However, within the same environment, you can also do analysis on climate-related data and climate risk related data. MATLAB allows one the freedom to derive insights from any type of input data, whether it be maps (including GIS files), images or big data. Take your analysis to the next level using machine learning algorithms, from decision trees to deep neural networks to forecast the damage of climate related events.

In this example, a machine learning model is implemented in MATLAB to predict the damage costs of various weather events based on location, time of year, and type, using historical data over 40 years, to train the model. In Figure 1, you can see an example of how different types of weather events are distributed across the United States (colour of bubbles) with the forecasted cost also indicated (size of bubbles). This type of data could be used to model the risk-associated investing to various companies based on say, geographical location, or proximity to natural disasters. This is only one type of example of how the climate data could be used.

 

 

 

 

 

 

 

 

 

 

 

Figure 1: Forecasted storm event damage

We encourage you to download the code and interact with the example – does this spark any ideas for climate risk analyses you might want to do? MATLAB can definitely be used as one integrated tool for modelling a plethora of risks and working with data across multiple industries and domains. The capability to model and estimate the financial risk that global warming poses could pave the way to creating scientific, actionable insights into financial risk caused by climate change.

What Can I Do Next?

Follow us

 
 

 

 

  • [1]  U. N. Environment, ‘Facts about the Climate Emergency’, UNEP – UN Environment Programme, Aug. 29, 2019. http://www.unenvironment.org/explore-topics/climate-change/facts-about-climate-emergency (accessed Jul. 06, 2020).
  • [2] ‘President Cyril Ramaphosa signs 2019 Carbon Tax Act into law | South African Government’. https://www.gov.za/speeches/publication-2019-carbon-tax-act-26-may-2019-0000 (accessed Jul. 07, 2020).
  • [3] E. Smith, ‘The numbers suggest the green investing “mega trend” is here to stay’, CNBC, Feb. 14, 2020. https://www.cnbc.com/2020/02/14/esg-investing-numbers-suggest-green-investing-mega-trend-is-here.html (accessed Jul. 09, 2020).