Title: Is it more rational to manage interest rate risk than weather risk?
1Is it more rational to manage interest rate risk
than weather risk?
- WRMA European Meeting 2006
- Toulouse, october 4th
- Pr. Didier Marteau Graduate Business School of
Paris, - Aon France
- and Ralf Holz (PhD), Aon France
2Is it more rational to manage interest rate risk
than weather risk?
- Its in fact a serious question that every
corporate should raise and be able to answer - In reality, few firms have been modeling their
weather exposure, and most of them allocate more
resources to managing financial risk than weather
risk. - Managers, economists, politicians pay a close
attention to the monetary policy, but are we sure
that a 25 basis point drop of the repo rate by
the European Central Bank has a bigger effect on
the economy than a 1Celsius change of the
average temperature in october ? - Some sectors are clearly more weather sensitive
than interest rate or currency sensitive (energy,
textile, food and beverage, leisure and tourism,
distribution, transport, construction) - So, how can we explain that corporates are still
reluctant to manage weather risk ? - Is it not time to replace traders by climate
experts and euro-dollar forecasting by weather
forecasting?
3How can we explain that corporates are still
reluctant to managing weather risk ?
1. Imperfect knowledge of weather risk compared
to financial risk
2. Modeling weather exposure is relatively
complex and hedging solutions are still
relatively new
4 basic raisons
3. Weather hedges are assumed to be
expensive
4. Weather risk management raises difficult
questions of corporate governance
4- 1. Imperfect knowledge about weather risk
compared to financial risk - 1.1 Comparative analysis of the volatilities of
weather variables and financial variables - 1.2 Economic analysis of the weather sensitivity
in some sectors
51.1 Comparative analysis of the volatilities of
weather variables and financial variables
- In the financial world, volatility is expressed
by the annualised standard deviation of asset
returns. - If the distribution of asset returns is normal
, the confidence interval associated to 1
standard deviation is 68, and the interval
associated to 2 standard deviations is 95. - Examples of historical volatility
- US stock index (01/1802 09/2006)
- Euro-dollar exchange rates (04/01/1999
09/25/2006) - 10 year T-Bond yield (01/1962 09/2006)
- Spot oil price (01/1987 09/2006)
- Carbon emission prices - EUA allowances (07/2005
09/2006)
61.1 Comparative analysis of the volatilities of
weather variables and financial variables
- US stock index 01/1802 09/2006 (logarithmic
scale) - The slope is quite constant, with an average
annual rate of return equal to 6.02 (continous
time)
Oil crisis 1973
71.1 Comparative analysis of the volatilities of
weather variables and financial variables
- 1 invested in 1802 in a stock index gives 216
010 in 2006. - 1 invested in 1802 in a T-bond index gives 464
in 2006 (3 continous time rate of return).
3 Bond Yield
81.1 Comparative analysis of the volatilities of
weather variables and financial variables
- US stock index 01/1802 09/2006 Distribution
of annual returns
91.1 Comparative analysis of the volatilities of
weather variables and financial variables
- Euro-dollar exchange rates (01/1999 09/2006)
101.1 Comparative analysis of the volatilities of
weather variables and financial variables
- Euro-dollar exchange rates Distribution of
daily variations (01/1999 09/2006)
Annualized volatility 8,6
111.1 Comparative analysis of the volatilities of
weather variables and financial variables
- Europe Brent Spot Price (01/1987 09/2006)
121.1 Comparative analysis of the volatilities of
weather variables and financial variables
- Europe Brent Spot Price Distribution of daily
price variations (01/1987 09/2006)
Annualized volatility 37
131.1 Comparative analysis of the volatilities of
weather variables and financial variables
- Carbon Emission Allowances Powernext spot
prices (07/2005 09/2006)
141.1 Comparative analysis of the volatilities of
weather variables and financial variables
- Carbon Emissions Allowances Distribution of
daily price variations (07/2005 09/2006)
Annualized volatility 75,8
151.1 Comparative analysis of the volatilities of
weather variables and financial variables
- US 10 year government bond yields (01/1962
09/2006)
Data Federal Reaserve
161.1 Comparative analysis of the volatilities of
weather variables and financial variables
- US 10 year government bond yields Distribution
of weekly variations (01/1962 09/2006)
Annualized volatility 12,3
171.1 Comparative analysis of the volatilities of
weather variables and financial variables
- Monthly Mean Central England Temperature August
(1800 2006)
C
Data Hadley Centre
181.1 Comparative analysis of the volatilities of
weather variables and financial variables
- Monthly Mean Central England Temperature August
(1800 2006) Distribution of
annual returns
Annual volatility 9,9
The volatility of temperature is at the same
level as the exchange rate volatility.
191.1 Comparative analysis of the volatilities of
weather variables and financial variables
- Monthly Mean Central England Temperature
November (1800 2006)
C
Data Hadley Centre
201.1 Comparative analysis of the volatilities of
weather variables and financial variables
- Monthly Mean Central England Temperature
November (1800 2006) Distribution
of annual returns
Annual volatility 34
Data Hadley Centre
Exactly in the same way as interest rates
volatilities, the volatility of temperatures
increases when the average temperature decreases
(the volatility of 3M interest rates on yen e.g.
is much higher than that of 3M dollar interest
rates).
211.1 Comparative analysis of the volatilities of
weather variables and financial variables
- Annual precipitation rate Paris (1800 2006)
Annual volatility 27
Data Meteo-France
The volatility of annual precipitation rates
is comparable to that of crude oil.
221.1 Comparative analysis of the volatilities of
weather variables and financial variables
- Daily maximum hourly averaged wind speed
(Vlissingen, Netherlands)
Data KNMI
231.1 Comparative analysis of the volatilities of
weather variables and financial variables
- Daily maximum hourly averaged wind speed
(Vlissingen, Netherlands)
Daily (!) volatility 37,1
Data KNMI
241.1 Comparative analysis of the volatilities of
weather variables and financial variables
- Two main conclusions
- The volatility of the weather variables is at
about the same level as the volatility of
financial indices. - Consequently, there is no reason to hedge
financial risk and bear weather risk except if
your firm is a weather firm or is less
sensitive to weather risk. But in many sectors,
this last assertion is clearly wrong
251.2 Economic analysis of the weather sensitivity
in some sectors
- Beer consumption and temperature
- Sensitivity relative change in consumption
per 1C change in temperature - At 25C 1C change signifies a 7 change in
consumption !! - R2 explained volatility /total volatility
0,80 - If 80 of change of beer consumption is
explained by a change in temperature, what does
really explain the financial results in beer
industry quality of management, level of
interest rates, marketing policy, coca-colas
innovations or weather ???
26The beer-temperature correlation is quite
instable geographically with two consequences
the need for defining precisely the local
exposure, and paying attention to correlation
risk if you cross-hedge.
1.2 Economic analysis of the weather sensitivity
in some sectors
Data Météo-France
271.2 Economic analysis of the weather sensitivity
in some sectors
- Weekly electricity consumption in France as a
function of temperature R2gt90
281.2 Economic analysis of the weather sensitivity
in some sectors
- Consumption of natural gas (distribution zone
around London) as a function of temperature (
wind speed correction) R2gt90
291.2 Economic analysis of the weather sensitivity
in some sectors
- Productivity in the construction sector as a
function of temperature and wind speed
301.2 Economic analysis of the weather sensitivity
in some sectors
- With data from the French Textile Industry Office
over the period 1977-2006, we studied the
correlation betwen textile sales and temperature.
For two seasons, spring and autumn, we obtain a
R2 close to 0.8, with a positive correlation in
spring and a negative one in autumn, especially
september
311.2 Economic analysis of the weather sensitivity
in some sectors
Beta relative change in sales for 1Celsius
change
Women buy clothes whatever the temperature
32How can we explain that corporates are still
reluctant to managing weather risk ?
- Modeling weather exposure is relatively complex
and hedging solutions are relatively new - Four basic differences betwen the financial risk
and the weather risk - 1. Weather risk is generally a volume risk, not
a price risk. If the euro-dollar moves up by 2,
your invoice in dollars moves up by 2. If the 1
year interest rate moves up 25 basis points, your
financial expenses move up by 0,25 nominal.
But if the october average temperature moves up
by 1C, your textile sales dont move down by 1C
!!! So internal or external expert resources are
required for analysing the weather sensitivity - 2. Weather risk is a strategic risk, not a
transactional risk. Transfering weather risk is a
strategic decision related to the identity of the
firm. Are Eurodisney or Parc Astérix weather
stocks or leisure stocks ? Does the equity
holder invest in weather or entertainment ? If
its in weather, the management has to bear the
weather risk as a source of risk and return, and
not to transfer it.
33Modeling weather exposure is relatively complex
and hedging solutions are relatively new
- From a theoritical point of view, temperature is
a new financial asset. - On one hand, the weather market is
incomplete , raising problems for pricing
weather derivatives. For instance, you cannot
duplicate buying forward by borrowing, buying
spot and carrying. Consequently, you cannot
price the forward temperature as - Spot temperature cost of financing carry of
temperature - The forward temperature will be traded close to
the expected spot temperature adjusted by a risk
premium, and not with the cash and carry
relationship. So, prices may be considered by
end-users as less transparent. - On the other hand, the asset price is not
affected by the continuous-time arrival of new
informations, as every other financial asset. It
implies for temperature a low volatility till 10
days before the beginning of the reference month,
followed by a high volatility, the short term
temperature forecasting horizon being an
obstacle for involving traders and reducing the
risk premium. -
34Modeling weather exposure is relatively complex
and hedging solutions are relatively new
Falling emission prices
UN approval for N20 CDM projects in Korea and
Brazil
35Modeling weather exposure is relatively complex
and hedging solutions are relatively new
36Contrat Chicago
Contract HDD January
Moyenne (10 ans). 1228
Moyenne (10 ans) 908
Moyenne (10 ans) 910
37Modeling weather exposure is relatively complex
and hedging solutions are relatively new
- 4. European end-users still suffer from a lack
of information about weather markets (how many
corporates know the number of European towns
traded on the CME?), and have not yet defined the
internal organisation and responsabilities for
hedging weather risk. The weather issue can be
taken in charge, either by the Treasurer
(technical skills about derivatives), the Chief
Financial Officer (strategic decision level), the
Risk Manager (in charge of managing the global
risk exposure) or even the marketing department
(hedging commercial risk). Weather risk
management is a cross-preoccupation.
38How can we explain that corporates are still
reluctant to managing weather risk ?
- 3. Weather hedges are assumed to be expensive
- Weather hedging is frequently considered as
expensive by end-users, and its cost
considered as an obstacle to the development of
the weather markets. Reality or perception?
39Why weather hedges are assumed to be expensive
- The issue here is that corporates are used to
hedge their financial risk on complete
markets , with prices built on arbitrage
strategies and easy to understand, whereas the
weather market is incomplete and appears less
transparent. - An asset is traded on a complete market if
its future pay-off can be duplicated by a
combination of other existing assets its the
case of most of forward contracts, options
contracts (Black-Scholes) In this case, the
asset price is simply the price of the
duplicating portfolio. This pricing methodology
is also called no-arbitrage pricing . - In a complete market, the asset price does
not depend on risk aversion and consequently does
not require any risk premium .
In a complete market, hedging looks as
free .
40Why weather hedges are assumed to be expensive
- You are a US importer and want to You are an
energy producer and want to - hedge your short 100 euros 3M hedge your
December exposure by buying - exposure by buying forward.
the DEC temperature forward. - Market data
- Spot euro-dollar 1.30 October temperature 15C
- US 3M interest rate 5 Euro 3M interest rate
3 - Euro 3M interest rate 3 Storage cost to be
determined (!) - Expectations
- Assume a binomial model, with Assume a binomial
model, with 2 forecast - 2 forecast values for the , values for the
temperature spot in - spot in 3M, either 1.36 or 1.24 december, either
8C or 12C - What is your forward price ?
41Pricing a forward contract on euro-dollar
(1.36 F)100
1.36
1.30
(1.24 F)100
1.24
spot
Pay-off of buying forward
- You duplicate buying forward by buying spot X
euros, borrowing X1.30 dollars at 5, and
lending X euros at 3. - The value of your portfolio at expiration is
following - X1.30(15/4) X(13/4)1.36 (1.36
F)100 - X1.30(15/4) X(13/4)1.24 (1.24
F)100 - 2 equations, 2 unknown variables, one unique
solution for forward pricing - (1 54)
- X - 100 / (1 3/4) and F
1.30 ------------------- -
(1 34)
42Pricing a forward contract on euro-dollar
- The euro forward rate does not depend on
individual expectations about the behavior of the
spot rate (you may replace 1.36 et 1.24 by any
values, you will find the same result), and is a
no-arbitrage price. It does not require any
risk premium. - Imagine the same pricing methodology for weather
- Your duplication portfolio consists in borrowing
euros, buying - october temperature spot at 15C, and storing it
(1 euro/month). - Your forward august temperature is 15C (1
310/12) 10 - euros ?????????????????????????
- Other models are needed, which take into
account the weather expectations, and include
logically a risk premium.For these two reasons,
the end-users, who are used to trade on
complete markets with a no-arbitrage pricing,
have more difficulties with weather.
43 Complete and incomplete
markets
Complete markets Forward price Spot price
cost of carry
Incomplete markets Forward price Expected
spot price /- risk premium
Reverse cash and carry faces limits
(convenience yield) oil prices
The asset cannot be carried power
The asset is not traded on the spot market
and cannot be carried weather indices
Hedge is costless no risk premium
Risk premium paid by the hedger
44Complete and incomplet markets the case of
options
- On complete markets (most of currency, interest
rates and equity markets), the forward price is
simply the spot price /- cost of carry. The
hedger does not pay any risk premium. - On incomplete markets, as the weather derivatives
market, the forward price cannot be extracted
from the cash and carry relationship (!). So, the
bank or insurer cannot hedge itself riskless, and
includes logically a risk premium in its price. - Ex if the bank quotes euro-dollar 3M, it can
itself hedge by borrowing dollars, buying euros
spot and lending euros. Its forward price will be
1.305. No risk premium is included in the forward
price. - But if the bank trades August temperature for a
client, it cannot hedge riskless by borrowing
euros (he can!), buying temperature spot (today
is 12C !), and carrying it till AugustThe bank
or insurer will hedge its contract through
diversification within a larger porfolio and
require logically a risk premium.
45Complete and incomplet markets the case of
options
- The magic idea of Black-Scholes (Journal of
Political Economy october 1973 Corporate
liabilities and option pricing ), is that buying
a call can be duplicated by borrowing and buying
delta stocks. Consequently, the price of an
option is the price of the duplicating portfolio.
No risk premium is required. A fundamental
assumption to this model is that the market is
complete. - If you ask for a call on temperature to your bank
or insurance company, it cannot hedge riskless by
borrowing and buying spot delta temperature.
Consequently, it will be obliged to use an other
model and include a risk premiumor to use
Black-Scholes and adjust the price because
riskless hedging is not possible.
46How can we explain that corporates are still
reluctant to managing weather risk ?
- IV. Weather hedge raises an important corporate
governance question
47Hedging or not hedging ?Four questions to raise
before answering
- Is the weather a strategic element of my
business? What is the pertinent market for a
theme park? Leisure market, culture
market,weather market? What is the relative
contribution of each risk factor, and especially
weather, to my earnings? - What do stockholders know about my weather
exposure? Do investors consider that they hold a
weather stock and do they expect a return for
that? Or do they consider that they bought a
leisure stock and dont want to bear any
weather risk? In any case, the management has to
communicate about its weather hedging policy.
48Hedging or not hedging ?Four questions to raise
before answering
- If the weather is not considered to be a
strategic element of business, and has to be
transferred, who is responsible for hedging
weather risk? Managers or stokholders? The
financial theory teaches that stockholders can
hedge themselves through diversification, and
that hedging a diversifiable risk is free and not
a managersjob. The only theoretical reason for
hedging weather, from the managements
perspective, is then avoiding default. But is
weather risk a specific risk, or a systematic
risk, and thus not diversifiable? - What do my competitors do? According to game
theory, if you hedge by your own, and if your
competitors dont hedge, you bear an additional
risk to have a lower financial result (because of
hedging losses) if the weather has a positive
impact on your business. Is the managements
performance evaluated on the basis of absolute
criteria or relative criteria? Do you hedge your
competitivity?
49Conclusion
- Weather risk is not restricted to extreme events
and is associated to the normal volatility of
weather indices. - Most economic sectors, and not only energy, are
weather sensitive, and have more reasons to
manage weather risk than interest rate risk - As the weather market is incomplete , a risk
premium has to be paid by the hedger, what
explains that weather hedging looks sometimes
expensive . - Weather hedging has to be implemented either by
the stockholders or by the managers, essentially
through diversification, bank and insurance
products, or financial markets. - Temperature, energy prices, carbon emissions
allowances prices, are tightly connectedShould
we manage globally these new environmental risks
in a Value at Risk approach or separately ?
50 The end Time is over and weather not cold
enough for a chocolate break. We go for an other
presentation.
Source Météo Franceobviously
51Illustration a simplified case study
- You are a beer producer and want to hedge your
weather risk in South of France over august. - Your sales budget in this region has been built
on a historical average temperature of 21C,
maybe corrected for a warming trend, and the
treasurer is in charge of protecting the 10
millions euros of expected earnings. - A correlation analysis showed that, in august, a
1C change in temperature led to an 6 decrease
in earnings. This linear approximation is assumed
to be correct between 18C and 24C. The
historical volatility in august is 10. - You decide to hedge your exposure on the CME
through the CAT futures contract. - Four questions
- What is the CAT futures contract?
- Do I sell or buy contracts?
- How many?
- What are the residual risks?
52What is the CAT futures contract?
- COMMODITY SPECIFICATIONS Cumulative Average
Temperature (CAT) - Each particular CME European CAT Index is the
accumulation of like daily average temperatures
over a calendar month. The accumulation period of
each CME European CAT Index futures contract
begins with the first calendar day of the
contract month and ends with the last calendar
day of the contract month. - The daily average temperature is defined as the
arithmetic average of the maximum temperature
(Tmax) and minimum temperature (Tmin), measured
at the following times for each location, as
reported by Earth Satellite Corporation. - Paris-Orly, France (WMO 07149) Between 0600 UTC
the current day and 0559 UTC the following day
for Tmax, and between 1800 UTC the previous day
and 1759 UTC the current day for Tmin - On october 4, 2006, the Paris CAT index August
2007 is supposed to be traded at 635, i.e an
average temperature of 635/31 20.5C - For simplicity, we assume that 1C is equals 20
euros
53Do I sell or buy contracts?
- Clearly, you.sell contracts
-
-
-
-
-
-
-
Temperature DOWN
54How many?
- Two steps
- First, we define the weather sensitivity of
earnings by the relative change in earnings
divided by the absolute change in temperature - dE/E
- S ------------
(1) - dT
- From (1) we derive dE S x E x dTe 6 x 10
000 000 x dTe (2) - Equation (2) is your exposure equation
- Then, you sell H (optimal ratio) futures
contracts at a price F, transforming your
exposure equation (2) in - dE ( 6 x 10 000 000 x dTe ) (H x
20 euros x dF ) (3) - or dE (6 x 10 000
000 x dTe ) (H x 20 euros x dFA x 31) - The optimal hedge ratio H can be deduced as
following
You sell contracts
Assumption your budget-temperature is the
traded future temperature
55How many?
- Scenario 1 the average temperature next august
is finally 18.5Cand beer sales are lower than
expected - The CAT contract expires at 573 (18.5 x 31)
- Exposure loss (compared to budget)
- 6 x 10 000 000 x (18.5C 21C) - 1 500 000
euros - Futures gain 968 x 20 x (635 573) 1 200
000 euros - Your physical loss is largely offsett by your
futures gain. - Why not totally? Because your sales budget was
based on your own expectation (21C) higher than
the forward traded temperature (20.5C). So dTe /
dF was not 1. - YOU CANNOT HEDGE YOUR OWN EXPECTATION, BUT ONLY
LOCK-IN THE FUTURES OR FORWARD PRICE . Thats
true on any market (foreign exchange, interest
rates).
56How many?
- If you are a US importer and want to hedge your 3
month short exposure, you can only lock-in a
1.31 euro-dollar (3 month forward rate) and not
the 1.30 spot rate. The euro-dollar premium is
against you and in favor of the US exporter. - If the yield curve is sharply increasing, you
cannot lock-in the spot interest rate, but a
largerly higher forward rate. - If the temperature is traded at 20.5C, you
cannot lock-in a 21C temperature indexed
budget. - ? Thats the reason why I recommend to use the
forward rate as the - budget exchange rate or budget interest
rate , and the forward temperature as the
budget temperature , because no treasurer is
able to hedge a different rate without taking
risks.
57What are the residual risks?
- Some corporates hesitate to hedge weather because
of correlation risk between the temperature
which affects their business and the traded
temperature. - Three answers
- The optimal hedge ratio takes obviously into
account the imperfect correlation between
physical and future temperature. - Correlation risk is not specific to weather, and
corporates face it regularly when they hedge
interest rate risk - Correlation risk is a residual risk, lower than
weather risk itself. - Banks and insurance companies can offer tailored
hedges. -
58What are the residual risks? Optimal hedge ratio
takes into account correlation risk
- Consider again the beer producer His risk is in
South of France and the CAT contract is related
to the Paris temperature - dE (6 x 10 000 000 x dTe ) (H x 20 x dFA x
31) - As dTe and dFA are correlated random variables,
we try to minimize the variance of dE and obtain
- 6 10 000 000
sdT - H ------------------------ x ? x ----
- 20 31 sdF
Meteo France delivers this information
(correlation coefficient)
59What are the residual risks? Optimal hedge ratio
takes into account correlation risk
- Example
- Correlation South of France Paris
- ? 0.75
- Temperatures volatility in Paris and South of
France - sdF 10
- sdT 12
- Optimal hedge ratio
- 968 x 0.75 x 12 / 10 871 contracts
- We call correlation risk the risk that the
correlation ? moves over the hedging period. But
the model takes into account the imperfect
correlation between Paris - and South of France by adjusting the hedge
ratio. - EVEN A BAD CORRELATION REDUCES THE WEATHER RISK
! - Theoretically, the reduction of variance of
the initial exposure is equal to the initial
variance times ?2 - ? 1 reduction 100
- ? 0.75 reduction 56
- ? 0.5 reduction 25
- ? 0 reduction 0
60What are the residual risks? Optimal hedge ratio
takes into account correlation risk
Data Meteo-France, Powernext