Bob Litterman is the former head of risk at Goldman Sachs worldwide and now a partner in New York based hedge fund Kepos Capital. Bob is co-designer of the Black-Litterman Quant model for portfolio allocation which is used extensively in the industry. Bob has been very active in the US and Canadian pension industry helping to brief funds on climate risk and how best to consider it in the context of their investments. Bob has been widely published on the subject and comes to Australia to share his understanding of how to better manage risk.
The details of economic models to price climate risk, and the calculations, are daunting, but humans have an incredible intuition. We can solve these problems in our heads in a fraction of a second. Think about a bicyclist racing down a mountain road. This is a true story. A friend of mine was racing down a beautiful, windy, forested, road outside of Palo Alto, when he came upon a traffic accident which was being cleaned up by the fire department. The road was completely blocked, and wet. From the time he saw this scene until he reacted was a fraction of a second. He grabbed the brakes as he formulated a plan to attempt to avoid a disastrous crash. There is an optimal amount to brake as you attempt to maintain your balance while obtaining maximum deceleration. We have the same problem to solve: what is the optimal amount to brake. My friend was unable to stop before hitting the wet pavement, losing control, and he broke is collarbone, but felt lucky it wasn’t worse. Perhaps think about a passenger train, a vehicle with much more inertia, like the passenger train that derailed in New York. That’s the mental picture you should have as you try to think about how hard to brake. That is, where to price climate risk. The price of GHG emissions is the brake.
The most important consideration in pricing climate risk is to recognize that there are two fundamentally different types of risk: diversifiable risk and non-diversifiable risk. Only the latter deserves a risk premium. Diversifiable risks should be priced at the level given by the discounted present value of future damages. Non-diversifiable risks deserve a higher price – they require a risk premium. Examples of non-diversifiable risks include: nuclear war, global pandemics, equity risk, and climate risk. Sometimes the market fails to charge an appropriate risk premium. Mortgages were thought to be fully diversifiable. They were mixed together, packaged up, and given a AAA rating, based completely on the faulty assumption that they were fully diversifiable. Not pricing this risk appropriately meant that many systemically important financial institutions loaded up on mortgage risk. They loaded their balance sheets with securitized mortgage obligations to such a degree that when the systemic risk was recognized, and mortgages were repriced appropriately, the financial crisis caused the great recession. There is an important lesson here: Not pricing risk appropriately leads to disaster.
Payoffs in bad states of nature are like insurance. Such payoffs occur in states with high marginal utility. They have a negative risk premium. Cash flows from such investments have discount rates below the risk free rate.
Physical scientists can give us an estimate of the distribution of temperatures over time associated with a particular path of emissions. What they can’t give us is any precise estimate of the capacity of the earths atmosphere to safely absorb emissions. Some estimate 350 ppm, this is the origin of the group 350.org. Others estimate 450 is safe, others 550. No one knows for sure. Of course humans have already increased the level of carbon dioxide in the atmosphere from 280 to 395 in a very short period of time. In currently is rising at a rate of 2.5 per year, and that rate is increasing over time. Think about that probability distribution. What is the probability that a long-run GHG level of 550 would lead to a catastrophe? What about 450? What about 350? The lower tail of that distribution is what we need to worry about. At what point does the reservoir overflow.
What climate modelers do is to use a low risk aversion.
States of nature versus intertemporal substitution
I was dabbling in the economic literature on pricing risk when I realized that the standard utility function that all the standard models used was inadequate to the task. Counterintuitively, increased risk aversion in the standard utility function caused the discount rate to increase, which caused the appropriate price to decline. It is actually a well known problem, which affects pricing equity risk as well, and was addressed 25 years ago in a more general utility function called Epstein-Zin utility. As you can see, because of this limitation, the standard approach was to use a very low level of risk aversion to maximize the price – but of course what that implies is that there is NO risk premium in the standard approach to pricing climate risk. This is not good! This was recognized by the US government. They were honest: their $21 estimate included no risk premium.
By the way, the equity risk premium historically has been puzzlingly high – over 500 basis points per year. Such a high level of risk aversion implies a very high price for carbon emissions.
Action on pricing carbon emissions is urgent.
There are two costs of delay. We are wastefully filling up the earths atmosphere. We are wasting a scarce resource pushing increased costs onto future generations.
But the main cost of delay is that we increase the unknown probability of a disaster. Remember the Johnstown flood. Strict liability refers to certain actions, such as filling a reservoir, which are inherently dangerous. Failure to have seen exactly what might go wrong is no excuse.
In fact, a higher level of risk aversion pushes the entire expected path of emissions prices up higher. Think of this forward price path as the expected level of braking over time as we deal with the resolution of uncertainty with respect to both climate impacts and technological advances.
Usually at this point in my talk I spend time explaining the implications for investors. There are implications for equity benchmarks, there are governance issues, and there are implications about market structure and price mechanisms.
Stranded assets will re-price to reflect changing expectations of forward prices, rather than changes in actual emissions prices. Assets like coal will be negatively impacted when expectations change, not when emissions are actually priced.
Aviation has promised: to create a “market-based measure” to reduce emissions The reality is that aviation seems to have no intention of creating appropriate incentives, or even to recognize that appropriate incentives – the social cost of carbon – should be created for the conservation of emissions But aviation needs capacity to create future emissions Unlike most emitters, commercial aviation will require the ability to create carbon dioxide emissions for the foreseeable future The atmosphere’s capacity to safely absorb emissions is a scarce resource that aviation needs, and will need in the future It is in aviation’s interest to lead the effort to create appropriate incentives for the entire global economy to conserve on emissions production today Owners of aviation shares have an important role to play Management often focuses too much on short term profits Long-term owners such as sovereign wealth funds and pensions have longer term priorities -- such as immediately creating appropriate incentives for the conservation of emissions
The Price of Climate Risk
Climate Change: Some questions
Is climate change real?
Is uncertainty about climate change real?
Is a devastating natural disaster outside the realm of
When, where, or how might a global catastrophe occur?
Does it matter how much carbon dioxide we put into the
Should adding emissions to the atmosphere be priced
What is the appropriate price for emissions?
Fossil fuel industry reaction:
A low carbon pathway would be too expensive,
thus none of our assets will become stranded
Think about dynamic optimization
With Uncertainty, Tipping Points And Nonlinear Responses
Where should climate risk be priced?
(economists call this: “the social cost of carbon)
There are 2 kinds of risk:
High risk aversion
Low risk aversion
The price of climate risk today
Diversifiable risk Expected damage
The Equity Risk Premium
US Historical Real Returns
Data are from http://www.econ.yale.edu/~shiller/data.htm
ERP = 4.75%
Stock real return = 6.4%
Bond real return = 1.6%
A consistent 475 basis points per year for the last 140 years
Equities pay off primarily in good states of nature
Consider a portfolio that pays off in bad states of nature
Data are from http://www.econ.yale.edu/~shiller/data.htm
An equally risky portfolio
long bonds and short equities earns
-310 basis points
What does the Equity Risk Premium have
to do with Pricing Climate Risk?
Pricing carbon emissions is a risk management
problem involving trade-offs between consumption
today and potential bad outcomes in the distant
This trade-off depends crucially on the degree
of societal risk aversion
Societal risk aversion can be calibrated to the
equity risk premium
Economic impacts depend on future
temperatures which are very uncertain
Science: 25 March 2012
Climate modelers generally use a low curvature in
the context of a standard CRRA utility function
Counter to intuition, in the standard utility function increasing the
risk aversion makes curbing emissions less urgent
Higher curvature has two impacts:
1) it increases the risk premium, but
2) it also increases the risk free discount rate
The second impact dominates and causes the price to decrease
Lord Nicholas Stern, for example, set a
degree of curvature that implies an
equity risk premium of around 12 basis
more than 30 times too low relative to
observed risk premia
Estimates of the social cost of carbon
from Anthoff, Tol, and Yohe (2009)
Increasing risk aversionWhy???
Higher curvature across states of
nature is required to fit the very
significant equity risk premiums
that we observe in the market
While lower intertemporal curvature
is required to fit the relatively
low risk free rates
that we observe in the market
Risk aversion Intertemporal substitution
Epstein-Zin utility can be calibrated to both
high risk premia and low interest rates
consumption ( time, states of nature ) consumption ( time, states of nature )
The rigidity of standard utility functions explains
why in most climate models increased
risk aversion lowers the price of emissions
The Appropriate Price for Carbon Emissions
Is Part of an Optimal Plan
The Appropriate Price
Trades off current consumption against future damages
Recognizes unknown impacts, and the potential for time compression and
Builds in a margin of safety
Anticipates risk reduction over time
Higher Risk Aversion
Increases the risk premium
Lowers the discount rate for future damages
Raises the price today and potentially lowers the expected future price
One cost of delay is higher future emissions prices
Another is increased risk of catastrophic outcomes
Higher societal risk aversion shifts the
appropriate emissions price path upward
forward prices will
be driven by the rate
of technological change
in emissions mitigation
Investors have exposure to emissions price risk
• Portfolio construction
• Tilt away from stranded assets e.g. coal and tar
• Appropriate, transparent business plan
assumptions about future emissions prices
• Hedging requires a forward market in
This recognition has implications for:
(any asset whose value will be negatively impacted by higher emissions prices)
Are they a risk or an opportunity?
Stranded assets will re-price to reflect changing expectations of
forward prices, rather than changes in actual emissions prices.
from CDP survey
current forward curve?
Stranded Assets Total Return Swap
¾ Coal index return
¼ Oil sands index return
S&P 500 index return
Stranded Assets Total Return Swap
Negative correlation to S&P 500 -.36
Annualized net total return 21.7%
1/3/2011 through 1/17/2014
S&P 500 15.9%
Tar sands 2.0%
A hedge which reduces portfolio risk
And adds a potential source of return
Better aligns investments with mission
Doesn’t impact underlying assets
Governance example: Aviation
• Aviation has promised:
– a “market-based measure” to reduce emissions
– but seems to have no intention to create appropriate incentives
• Aviation will need capacity to create emissions
– requires high energy content of liquid fuel for takeoff and ascent
– atmosphere’s capacity to safely absorb emissions is limited
– thus aviation has a special incentive to lead on this issue
• Owners of aviation shares have an important role to play
– management often focuses too much on short term profits
– long-term owners have longer term priorities, such as creating
appropriate global incentives to reduce emissions