Introduction

The management of weather risk has moved to the forefront in recent years as
corporate risk managers have sought to take greater control of risks that impact
their financial operations. Weather risk management reduces or eliminates a
company’s financial exposure to temperature, precipitation and other forms of
non-catastrophic weather. Awareness that weather risks can be managed has led
to increased participation by different types of providers and end-users, giving
the market greater risk capacity and momentum. With a meaningful segment of
the U.S. and global economies exposed to the effects of varying weather (certain
estimates suggest that between 10 and 20 percent of the U.S. economy is affected
by the weather) and with growing recognition by companies that exposure to temperature,
precipitation and other weather events can be managed, the weather risk market
is emerging as an important risk transfer mechanism.

Market Fundamentals

The market for weather risk management — which centers on protecting financial
operations against economic losses due to variations in weather variables such
as temperature, precipitation and snowfall — has evolved at a rapid pace
since the first deals were arranged in the U.S. market in late 1997. The first
U.S.-based temperature transactions, embedded in over-the-counter power contracts,
were created by utilities in late 1997. By early 1998, stand-alone temperature
transactions — arranged as independent financial transactions rather than
as components of power deals — began appearing in a growing number of U.S.
locations, and by late 1998 had expanded to include various European reference
locations. Activity in the Japanese market soon followed. The Chicago Mercantile
Exchange began listing and trading standard futures and options contracts on
10 U.S. temperature indexes in September 1999,¹
and the first temperature-linked bond appeared in the market shortly
thereafter. In July 2001, the London International Financial Futures Exchange
(LIFFE) announced the launch of temperature indexes on three European cities
(London, Paris and Berlin) and stands ready to introduce derivative contracts
should client interest develop.

Weather risk protection can be written on any one of several variables, including
temperature, precipitation, snowfall, streamflow²
and wind. These are all considered non-catastrophic forms of weather. Non-catastrophic
weather events, which have a reasonably high probability of occurrence, stand
in contrast to hurricanes, tornadoes and windstorms, which are classified as
low-probability catastrophic events. Catastrophic weather events are dealt with
through other forms of risk mitigation, such as catastrophe insurance, and attract
different types of participants. This paper’s focus is limited to the non- catastrophic
weather market.

Most activity in the market up to the present time has been in the form of temperature
contracts. In the U.S. market, temperature contracts are typically denominated
in Heating Degree Days (HDD) or Cooling Degree Days (CDD). An HDD measures the
relative coolness of a given location over a period of time. Specifically, an
HDD is the greater of (a) zero or (b) 65^o
Fahrenheit minus the day’s average temperature (calculated as the average high
and low temperatures from midnight to midnight); the HDDs for an entire period
are computed by summing up each day’s HDDs.³
CDDs, which measure a location’s relative warmth, are computed by selecting
the greater of (a) zero or (b) the day’s average temperature less 65^o
Fahrenheit. While HDDs and CDDs are very common in the U.S. market, other temperature-related
structures are emerging. For instance, the critical day contract, which measures
temperature on a single day (or group of days), and provides protection if the
actual temperature is above or below a specified level, is becoming increasingly
popular. The European temperature market is geared primarily towards use of
peak or average temperatures, rather than HDDs or CDDs.4
Contracts based on rainfall or snowfall are generally measured in terms of inches
or centimeters of precipitation falling during a pre-defined time period at
an approved weather station location. Likewise, wind protection is based on
peak or average windspeeds recorded by an anemometer at a given location.

To date, most deals in the weather market have been written in the form of over-the-counter
derivatives (as OTC put and call options, collars and swaps), or insurance and
reinsurance contracts. Only one weather bond transaction has been launched and
turnover in the listed derivative market (via the Chicago Merchantile Exchange
contracts) has been extremely light. Whether LIFFE will have greater success
with its offering remains to be seen.

Identifying Weather Risks

In order to manage weather risks effectively, it is important to first identify
and quantify an institution’s exposure to weather variables. Only then can an
appropriate risk management solution be created. For instance, an electric utility
in the Southeastern United States might realize that it is exposed to a financial
risk from cooler than normal summers. When summer temperatures are mild —
perhaps 10^o or 15^o
Fahrenheit cooler than normal (where “normal” is defined as a long-term average)
— the utility’s residential and commercial customers might not run their
air-conditioning units. This results in less power being consumed by the customers,
and translates into lower revenues for the utility. While the utility can curtail
its power generation and lower its variable costs, it can do very little to
reduce its fixed costs. Once the risk has been identified (in this case an exposure
to cooler than normal summer temperatures), the impact on financial operations
needs to be quantified. This is an important — and sometimes complex —
exercise that requires estimating the sensitivity of the institution’s revenue
base to changes in temperature. Since utilities typically keep detailed records
of historical customer demand (by the day, hour or even minute), it is possible
to construct a function that matches historical demand with historical temperatures.

However, compiling data related to power demand is only half of the equation.
In order to create the correct risk function, it is crucial to have access to
relevant and accurate historical temperature data. These data series are often
available commercially. Compiling this data is not always as simple as it seems.
For instance, the location of the official temperature readings must be sufficiently
close to the utility’s base of operations to be relevant (or else a certain
amount of basis risk⁵ will arise). If there is not a
proximate weather station capable of providing accurate data, opportunities
to implement a risk solution can be severely limited. In addition, a sufficiently
long history of data (generally 30 to 50 years of daily observations) must be
compiled in order to allow proper valuation of the risk mitigation structure.
Finally, the data must be “cleaned” — missing values must be properly filled
in, changes in station location must be accounted for, trends created by urban
heat island effects and other phenomena must be identified and resolved, and
so on. The data compiling portion of the risk evaluation process is so important
that the point bears repeating: a sufficient amount of high quality, historical
data is a prerequisite for the construction of any weather risk solution, regardless
of the underlying variable or region.

Risk Management Solutions

Once the risk/revenue function has been constructed, an appropriate risk management
solution can be created. Depending on the overall goals and strategies of the
company, it may choose to purchase an option, sell an option, purchase a collar
or some other multiple option position, or enter into a swap. For instance,
if the utility is primarily concerned about a cool summer, it may choose to
buy a CDD put option, which provides an economic payoff if actual CDDs fall
below the pre-defined CDD strike. If the utility doesn’t want to pay a premium
to obtain this protection and is willing to forgo revenues in the event temperatures
are higher than normal, it might enter into a zero-cost collar by purchasing
a CDD put and selling a CDD call. This would enable it to obtain downside protection
(i.e., against the cool summer), and pay away upside revenues (i.e., from a
hot summer), with no premium payment. The utility might choose to do so as higher
temperatures (e.g., higher CDDs) translate into more customer usage of air conditioning,
greater power consumption and larger revenues. As the utility earns incremental
revenues, it is prepared to make a payment on the short call option. Other risk
management structures can, of course, be created.

For example, suppose the ABC Utility, operating in the Miami region, is attempting
to protect its revenue base from a cool summer. Based on analysis of historical
temperature and power demand data, it determines that it loses revenues when
CDDs (covering June 1 to September 30, a typical CDD period) fall below 5,000,
as measured at the Miami International Airport. Analysis of its revenue function
indicates that for each CDD below 5,000, ABC Utility loses approximately $10,000.
Accordingly, it purchases a CDD put from a protection provider at $10,000/CDD,
to a maximum payout of $3 million.⁶ At the
conclusion of the contract, the CDDs recorded at Miami International Airport
amount to 4,775. Based on the CDD put, ABC is entitled to receive $2.25 million.
Though ABC recorded lower revenues as a result of lower power use by its customers,
it offset the loss through a gain on the CDD put. Had CDDs been in excess of
5,000, ABC wouldn’t have received a payout on the put, but would have earned
greater revenues from its power business. By using a CDD put, the utility mitigated
its exposure to weather risk and removed a source of uncertainty from its financial
operations.

This example illustrates only one of many strategies that can be used in the
weather risk market. Companies from many industries can be exposed to weather
risks. For instance, construction companies, theme parks and beverage companies
might be negatively impacted by rainy or cold weather (e.g., rain and cold temperatures
can result in construction stoppages, low theme park attendance and reduced
purchases of cold beverages); a commercial vineyard might be negatively impacted
by an excess or lack of rain and overly hot or cold temperatures (e.g., grape
production can suffer as a result of any/all of these adverse conditions); and
a ski resort can be negatively impacted by lack of snowfall and hot temperatures
(e.g., low attendance if conditions aren’t conducive to skiing).
In addition to the quantitative aspects of the risk process described above,
corporate considerations play a role in determining an optimal risk management
solution. As indicated earlier, a company can obtain the protection it desires
from either derivatives or insurance. Each instrument has unique features that
appeal to different participants in the marketplace. For instance, certain corporate
institutions have a long history of purchasing insurance and are more comfortable
buying protection in the form of a standard insurance policy, even though the
economic payoffs may be identical from using a derivative.

Some reasons why a company may prefer to use insurance are because policies
are regulated by state (and national) authorities and are scrutinized closely.
In addition, the institution might derive certain tax and accounting benefits
from using insurance. For instance, under certain scenarios insurance premium
payments can be tax deductible.

In contrast, some companies prefer the ease and flexibility of derivative structures,
which can be arranged, documented and executed quickly. Companies using derivatives,
however, are required to adhere to Financial Accounting Standard 133, which
requires derivative payables and receivables to be reported through the balance
sheet; FAS 133 effectively imposes additional monitoring and reporting requirements
on derivative users.

Growth Trends

From rather modest beginnings in 1997, the market for weather risk protection
has grown rapidly. In 2001, the Weather Risk Management Association engaged
PricewaterhouseCoopers to conduct a survey of weather risk contracts executed
among Weather Risk Management Association members between October 1997 and March
2001.⁷ The survey revealed that approximately 4,800 weather
derivative contracts, with a notional value of $7.5 billion, have been transacted
since 1997. These findings indicate impressive growth in a relatively short
period of time. Figures 1 and 2 illustrate the growth trends in both contract
and notional terms.

Figure 1 – Growth in Weather Contracts, 1997-2000 (Number of Contracts). Source:
PwC Weather Risk Management Industry Survey.

The survey also indicated that over the past four years activity in the weather
risk market has been heavily concentrated in U.S. HDD and CDD temperature contracts.
For instance, in winter of 2000, 90.8 percent of notional contract value was
based on HDDs and 1.9 percent based on other temperature measures. Only 6.6
percent of contracts were written on rainfall and 0.2 percent on snow. A similar
composition was in evidence for summer of 2000, where 91.1 percent of notional
contract value was based on CDDs, 5.5 percent on other temperature measures,
1.5 percent on rain and one percent on wind.⁸
Winter protection remained dominant throughout the survey period.

Figure 2 – Growth in Weather Contracts, 1997-2000 (Notional Value). Source:
PwC Weather Risk Management Industry Survey.

From a geographic perspective, most weather risk transference activity has
occurred in the United States, though there is a gradual trend towards offshore
dealing. In 2000, 95.2 percent of winter notional contract value and 97.4 percent
of summer notional contract value was centered in the United States. This finding
is not surprising, since the market originated in the United States, and the
U.S. energy industry has been the earliest promoter of the market and products.
Despite the dominant position of the United States, there is some indication
that activity in the European and Asian markets is increasing, albeit gradually.
Tables 1 and 2 summarize the geographic trends since 1997-1998.

Table 1 – Geographic Distribution of Winter Contracts, 1997-2000. Source:
PwC Weather Risk Management Industry Survey.

Table 2 – Geographic Distribution of Summer Contracts, 1998-2000. Source:
PwC Weather Risk Management Industry Survey.

Within the U.S. market itself, geographic dispersion has increased since 1997-1998.
In 1997 winter contracts were based heavily in the Midwest (accounting for 72
percent of notional value), while in 1998 summer contracts were based largely
in the Midwest and East (which together accounted for 89 percent of notional
value). By 2000, the market featured much greater balance, with no particular
region accounting for more than 40 percent of notional value in either the winter
or summer. Figures 3 and 4 illustrate the geographic mix of U.S. weather contracts.

Figure 3 – U.S. Weather Contracts, Notional by Region, Winter. Source: PwC
Weather Risk Management Industry Survey.

Figure 4 -U.S. Weather Contracts, Notional by Region, Summer. Source: PwC
Weather Risk Management Industry Survey.

Conclusion

The market for weather risk management has emerged as an important mechanism
for managing institutional exposure to variables such as temperature, rainfall
and snowfall. Standard or customized risk management programs can be created
to mitigate a company’s exposure to the uncertainties of weather. Since weather
risks can be managed, it is no longer suitable for CFOs and treasurers to attribute
lost revenues or increased expenses to unusual weather. Just as these individuals
assume corporate responsibility for managing other dimensions of their financial
and non-financial risks — including exposure to interest rates, equity
prices, currencies, commodity prices, and so on — they can now assume responsibility
for managing their exposure to weather. Growing recognition of this fact is
fueling activity in the marketplace — adding liquidity, transparency and
choice to those opting to participate.

Footnotes

1 The CME reference indexes include Atlanta, Chicago, Cincinnati,
Dallas, Des Moines, Las Vegas, New York City, Philadelphia, Portland, and Tucson.
2 Streamflow is a measure of the amount of water flowing in a river and is of
major importance to utilities that generate electricity through hydroelectric
turbines. Low levels of streamflow (caused by insufficient precipitation and/or
cold temperatures that don’t permit snowcaps to melt) can impair a hydroelectric
utility’s ability to generate power and deliver electricity to customers.
3 For instance, if the average daily temperature in a location is 44Þ Fahrenheit,
the HDDs for that day amount to 21.
4 The new LIFFE indexes are based on average daily temperatures applied to a
monthly mean.
5 Basis risk arises as a result of imperfect correlation between the reference
location and the actual location that is sensitive to weather changes; for instance,
if the nearest weather station is 100 miles from a company’s base of operations,
changing weather patterns over the 100 miles can yield different readings at
the two locations — basis risk in this case is said to be significant and
the desired economic protection might not be achievable.
6 It is common practice in the weather market to include a maximum payout amount
on every contract; this amount is also referred to as ‘notional value,’ as discussed
later in the paper.
7 Though the WRMA/PwC survey included most of the major corporate, bank and
insurance participants active in the market, it was not inclusive of every institution
that has arranged or executed a weather risk transaction; accordingly, actual
activity totals may be understated by some margin.
8 PwC Weather Risk Management Industry Survey, June 2001.