Economic scenario generators can be a useful tool in modelling and forecasting, and can also be used to effectively reflect the risks associated with climate change.

In this episode, Senior Economist Sohini Chowdhury joins us once again to explain how economic scenario generators can be used by insurance companies and pension plans, and what impact climate can have on these scenarios. (In English)

Read the transcript

Les générateurs de scénarios économiques peuvent être un outil utile lorsqu’il est question de modélisation et de prévision, et peuvent également être utilisés pour représenter les risques associés aux changements climatiques.

Dans cet épisode, l’économiste principale Sohini Chowdhury se joint de nouveau à nous pour expliquer comment les générateurs de scénarios économiques peuvent être utilisés par les sociétés d’assurance et les régimes de retraite, et quelles sont les répercussions climatiques sur ces scénarios. Cet épisode est disponible en anglais seulement.

Lire la transcription (en français)

Welcome to Seeing Beyond Risk, a podcast series from the Canadian Institute of Actuaries. I’m Chris Fievoli, Actuary, Communications and Public Affairs at the CIA.

Scenario testing is an integral part of most actuarial work, and users have come to rely on a variety of generators to develop these scenarios.

Today, we’re going to take a closer look at economic scenario generators, which have applications in insurance and pensions, and which also have the ability to reflect the impact of climate risks.

Joining us once again to discuss this is Sohini Chowdhury, a Senior Economist with Moody’s Analytics. Welcome back to the podcast and thank you for joining us today.

Thank you, Chris. Thank you for having me.

Let’s start right from the top. Tell us, what is an economic scenario generator?

An economic scenario generator is – at its heart – a collection of several stochastic financial models of risk drivers and asset prices. These models generate multiple future paths, which are called scenarios for economic risk factors, such as interest rates, credit spreads, equity volatilities, exchange rates, inflation, etc.

These risk factor projections are then used to produce multiple future projections of asset returns, which depend on them; for example, corporate bond returns, equities, foreign assets, commodities, etc., and also flexible tools within the software.

The economic scenario generator software allows users to calibrate these models’ parameter sets, which could be the mean, the standard deviation of the distribution, etc., to a range of data and assumptions.

So that is what it is – it’s a collection of models which produce stochastic outcomes for financial risk drivers and asset prices.

Now let’s take a little step further. Why do insurance companies, pension funds, and asset managers need an economic scenario generator?

These businesses need an economic scenario generator for any application that requires an analysis of the full distribution of future paths of market risk variables because the future is uncertain. These applications can be broadly divided into two groups.

The first includes real-world applications like economic capital calculation, ALM, asset-liability management, stress testing, investment strategy design, and portfolio construction. For these applications, the aim is to produce realistic future paths and distributions of economic risk factors that sort of reflect forward-looking views.

The other kind of application where an economic scenario generator comes into play is risk-neutral applications, such as valuation of complex liabilities, especially with uncertain cash flows, where analytical approaches don’t produce solutions.

Under these applications, you would generate a distribution of future discounted playoffs, and the average of these playoffs would provide the best estimate of the liability value at today’s date.

So, under risk-neutral applications, you would be answering questions like, “What is the fair price for a product?” “How should the charge for any financial guarantee be invested in order to minimize the ALM mismatch?” Stuff like that. And risk-neutral applications also include valuing derivatives, hedges, and their sensitivities.

But why do we need an economic scenario generator to do this? Why can’t we just do these distributions and produce them using a simple stochastic process?

Since the behaviour of different asset prices and risk factors are correlated, we need a structural model to capture these dependencies between asset classes. This is important. Interest rates, equities, and credit spreads don’t just behave independently, they are all connected.

For example, nominal risk-free rates form the core of most of these models, and then risky asset returns are modelled in excess of these. You’re basically modelling the risk premia.

The structure also embeds some well-understood economic principles. For example, real interest rates are linked to nominal interest rates via inflation, and credit and equity models are linked via intra-sector correlation to capture the relationship between equity market returns and bond transition and defaults.

Basically, you need this structure to capture this interdependency that you see in the real world. One very good example which shows the advantage of a structural model is the following:

In a good economic scenario generator, the equity model has a factor modelling structure, which basically allows the modelling and calibration of hundreds of equity-type assets in a scalable and efficient manner.

The exposure of each equity asset to each of the factors drives the level of correlation between equity markets, and also captures the non-linear dependencies between the assets in the tail of the distribution; therefore, the height and correlation and dependencies between equity markets that are seen in times of stress is able to be captured.

There’s something that we call “tail correlation” – during a recession, the correlation between the S&P 500 and the FTSE, and all of these, they increase, and we’ve seen this all the time.

A non-structural model, where things are independently modelled, will not be able to capture this increased correlation during times of stress, whereas, in this particular example, a factor modelling approach is able to capture this.

Let’s bring climate into the discussion. It’s obviously an important topic, but can you just recap for us why climate change is an important issue for insurance companies, pension funds, asset managers, and other groups?

Climate change creates new risks and opportunities for asset managers, insurance companies, and pension funds by changing the risk and return profile of long-term assets and liabilities; these businesses must identify the risks and opportunities from climate or ESG [environmental, social, and governance], quantify these, and also report the results to meet the evolving regulatory disclosure requirements.

These can be summarized into the “four R’s.” The first R is regulation, which is evolving, of course, and depends on the industry and region. The second R is reputation, which is ensuring that your company is operating in line with net-zero and sustainable strategies.

The third R – very important – is risk management, ensuring a consistent view of risk across investment capital, liquidity, underwriting, and product design. So “how exposed are you?”, “how exposed is your portfolio to climate risks?” So, sort of like stress testing.

And the fourth R, which is the latest addition to this structure of ours, is revenue. Revenue is the one positive R; the first three R’s are negative – they are all risks. The fourth R is revenue, which is to be able to benefit from opportunities that climate change presents. So: regulation, reputation, risk management, and revenue.

Here is an example: a life insurance company would want to know how climate risks will impact the cash flows from the CRE [commercial real estate] assets they hold 20 to 30 years into the future. An asset manager will need to consider both the risks and opportunities from climate when constructing a future portfolio. This is called climate-aware investing.

How can an economic scenario generator incorporate these climate risks?

Since we don’t know how these greenhouse gas emissions, temperature, or climate-related policies and technologies will evolve in future, we need to rely on “what-if” scenario analysis.

Climate pathways are hypothetical scenarios which assume different future development of emissions, warming, and climate-related policies. Such scenarios are published by some regulatory bodies like the Bank of England and also by networks like the Network for Greening the Financial System [NGFS], which is basically a group of central banks across the world.

These bodies publish, publicly, on their website, climate scenarios, which is basically a hypothetical scenario where there is certain “X” degrees of temperature rise, wide degrees of greenhouse gas emission, and some technological development, carbon taxes, etc.

But the real challenge for financial institutions is to translate the climate conditions in these scenarios to the financial variables that are ultimately needed for better ALM analysis, risk management, or disclosures. In other words, to quantify the financial impact from climate change.

There can be a scenario where temperatures are four degrees higher than what they were in the pre-industrial age and a bunch of other parameters, but what does that mean for interest rates? Because unless I know what it means for interest rates, I can’t really say what it means for my portfolio. So that’s the gap.

And this is where an economic scenario generator comes in. How? Because it plays the crucial role of quantifying the financial impact of climate change. So, for example, the economic scenario generator will generate the future path of short and long rates, credit spreads, equities, and other asset prices (all the stuff that we talked about in the beginning) under a scenario in which the world attempts to transition to a low-carbon economy. But in a certain and disruptive manner.

This is called a disorderly transition or “late policy action scenario”; a scenario which is defined by high carbon taxes in 2030 to compensate for the lack of timely action on climate policies. The carbon taxes in this scenario push up inflation, so inflation is high, and depress real returns. So that is what you’re looking for at the end of the day.

These bodies, whether the Bank of England or the NGFS, publish a late policy action scenario, but (as for the gap) the economic scenario generator helps translate the conditions under those scenarios into financial variables by saying that, OK, your real returns for the different sectors will go down by “X”% or “Y”%, inflation will rise by “Z”%.

The financial impacts of other climate scenarios can also be similarly analyzed. For example, another scenario that is commonly analyzed is a “hothouse” world scenario or a scenario where there is no policy action.

This scenario features complete inaction on the part of policymakers, and therefore no transition risk on financial forms – it basically means that policymakers take no further action beyond what has been already announced.

And so, farms have zero transition risk. Transition risks are permanent shifts, which are driven by changes in policies, technology, carbon pricing like there are no carbon taxes. However, the absence of climate policies and regulation means that physical risks from a warming atmosphere are super high in this scenario.

So, you have immense physical risk from climate change, although no transition risk. But this is a very different scenario, so you need to be able to figure out how your portfolio reacts under each of these different hypothetical scenarios, simply because it’s all uncertain. We don’t know what will eventually happen.

The economic scenario generator really helps to quantify the financial impacts of climate change so that insurance companies, pension funds and asset managers can take more informed decisions and be better prepared for the future.

That was a really good introduction to the topic, thank you for that. And thanks once again for joining us today.

Thank you, Chris.

We now have over 100 episodes in our podcast series over the past three years, so we encourage you all to subscribe and you can do so through whatever platform you use to get your podcast content.

We would like to hear from you as well. If you have any suggestions or episode ideas, you can send them along to podcasts@cia-ica.ca.

And we’re always looking for content to put in our Seeing Beyond Risk blog. If you have some ideas you would like to share, you can contact us at seeingbeyondrisk@cia-ica.ca.

Until next time. I’m Chris Fievoli and thank you for tuning in to Seeing Beyond Risk.