A Report submitted in partial fulfllment of the requirements of the
course Derivatives Management Under the Guidance of Dr. Alok Pandey PGDM (International Business) 2!"# Animesh $erma (!IB"!) %Pratee& 'inha (!IB"())% 'udarshan Bhutra (!IB"**) +,-./-.' INTRODUCTION.........................................................................................................................................................3 WEATHER MEASURES.................................................................................................................................................4 HDD and CDD......................................................................................................................................................4 WEATHER DERIVATIVES STRUCTURE..........................................................................................................................5 Call and Put Options.............................................................................................................................................6 Weather Swaps.......................................................................................................................................................7 Collars (Fences)....................................................................................................................................................7 PRICING OF WEATHER DERIVATIVES...............................................................................................................9 PRICING MODELS FOR WEATHER DERIVATIVES..........................................................................................................9 1. BLACK-SCHOLES AND WEATHER DERIVATIVES................................................................................................12 2. SIMULATIONS BASED ON HISTORICAL DATA OR BURN ANALYSIS..............................................................14 . MONTE CARLO BASED SIMULATIONS..............................................................................................................15 WEATHER DERIVATIVES IN INDIA....................................................................................................................16 HEDGING WEATHER RISK USING WEATHER DERIVATIVES....................................................................23 INTRODUCTION The weather derivates market started in 1997 in the US as the frst transaction in this regard was recorded there but now this weather derivatives market has spread to all major markets. According to US energ market! US" 1 trillion o# the US econom is a$ected b the weather risk. % the ear &''' about US" (.) billion worth o# weather derivates were traded in the US. There are * major plaers in the weather derivates market + ,A-./T ,A./-S %-0./-S 12SU-A23/ A24 -/12SU-A23/ 30,5A21/S. /24 US/-S such as 6as and power marketers and utilities etc. 4erivative is a contract or a securit whose value or pao$ derives #rom the price o# an underling asset. 4erivatives help an investor to control the risks o# changes in the prices o# the underling asset. 7or eg. An e8porter who receives his paments in #oreign currenc is e8posed to currenc risk ie the risk o# home currenc appreciating with respect to the #oreign currenc. 9hen -upee started appreciating against the US 4ollar the te8tile e8porters and 1T companies #aces heav losses i# the had not hedged their positions. Weather Measures 9eather measures are considered underling assets o# the weather derivatives! as the price o# a #utures contract is an asset #or an option on a commodit. The two most common weather measures are : ;eating 4egree 4as <;44= or 3ooling 4egree 4as <344= : depending on the specifcs o# the contract. 1t is estimated that 9>?99@ o# the weather derivatives are now using these temperature parameters. 0ther measures are based on precipitation! which can be measured b the amount o# rain over a given time period or on Snow#all! measured b the amount o# snow <or sleet= over a given time period. HDD and CDD These weather measures are used to measure the demands that arise due to the departure o# the average dail temperatures #rom a base level. An ;44 <or 344= is the number o# degrees the daAs average temperature is above <or below= a base temperature. The are calculated as #ollows+ Daily HDD = Max (0, base temperature daily average temperature) Daily CDD = Max (0, daily average temperature base temperature) 9here! Base temperature is defned as! the pre?defned base temperature! and! Daily average temperature is measured as the average between the dail high and the dail low. To calculate the accumulated ;44s <or 344s= over a specifed time period! a simple addition o# the dail ;44s <or 344s= is per#ormed. A #ew other measures used in 9eather derivatives are+ /nerg 4egree 4as </44s=! measured as the sum o# ;44 or 344 #or each da! 6rowing 4egree 4as <644s= defned as the degrees between a certain range. 644s are used o#ten in agriculture. Weather Derivatives Structure ,ost weather derivative trading are either swaps! or call and put options or a combination o# these. 3ustomiBed structures have started coming up based o specifc needs! like binar or digital options. These either pa a f8ed sum or Bero depending on whether the pa?o$ is satisfed. 4ouble trigger options are another e8ample! which pa?o$ onl i# the two conditions are met. Pay-! = "pe#i$ed dllar amu%t & ('"tri(e) HDD r CDD level a#tual #umulative HDD r CDD level) 3ontracts are usuall capped! i.e. onl a ma8imum amount o# paout can change hands. This is done so as to limit the ma8imum amount o# paout b an o# the counterparties. Call and Put Options As mentioned earlier! ;44s and 344s act as the underling asset #or the weather derivatives. Since weather is not a tradable asset! a dollar value is linked to ever degree da in the pa?o$ calculation. 5a?o$s o# the weather puts and calls calculated as+ Pay-! Call* p (+ , DD) &(Max(0, - t./ 0) Pay-! Put* p (+ , DD) &(Max(0, 0 - - t./ ) where! p <" C 44= is the per degree pa?o$! D tET is the underling ;44 <or 344=! and . is the strike <in terms o# the underling measure= An investor! who has purchased <is long= the call option! will receive the pa? o$ i# the recorded ;44 or 344 #or the season are greater than the strike .. An investor who has purchased the put option on the other hand will receive the pa?o$ i# the ;44 or the 344 are lower than the strike. +all and put options 0ith a ma1imum pa2"o3 (cap) The reason a cap is specifed on the call and put options is to avoid e8cessive pa?o$s. The pa?o$s are then defned as! Pay-! #all* Mi% (p (+ , DD) &(Max(0, - t./ 0), h) Pay-! put* Mi% (p (+ , DD) &(Max(0 - 0, - t./ ), h) where h is the ma8imum pa?o$ in dollars. Weather Swaps A swap is a combination o# put and call options! which have the same strike and are on the same underling location. -evenue stabilit can be provided b degree da swaps. Pay-! s1ap* FMi% (p (+ , DD) &(Max (0, - t./ 0), h)2 3Mi% (p (+ , DD) &(Max (0 - 0, - t./ ), h)2 An investor who is long the swap! will receive paments! i# the recorded ;44 or 344 are greater than the strike! and will make paments! i# the recorded ;44s or 344s are lower than the strike. Collars (ences! A collar is a spread position that insulates the buer #rom e8treme movements in the underling asset. 1t consists o# purchasing an 0T, call <or put= with a particular strike! and fnancing this with the sale o# an 0T, call <or put= with a di$erent strike. Pay-! #llar* FMi% (p (+ , DD) &(Max (0, - t./ 0 4 ), h)2 3Mi% (p (+ , DD) &(Max (0 5 - 0, - t./ ), h)2 PRICING OF WEATHER DERIATIES 9hen the trading o# weather derivatives started initiall in 1997 there were a ver #ew participants in this market and there were huge bid ask spreads but currentl when the number o# participants has increased signifcantl this bid ask spread has reduced drasticall. The #ollowing table mentions the measuring stations and ticker smbols o# #utures contracts that are traded in 3hicago ,ercantile /8change. Pricing M!"e#s $!r Weather Derivatives 0ne can price weather derivatives using one o# the man was available. Some models #ocus on the ;44 and 344 directl. The problem with this approach is that a#ter we calculate the weather measure b modeling ;44 or 344 directl! a lot o# in#ormation is lost as the values o# ;44 and 344 can be Bero also. Some models #ocus on temperature directl and then e8tract the ;44 and 344 #or each temperature scenario. This method is a better and a more comprehensive method. The e8ample mentioned below clearl shows what in#ormation can be lost i# we decide to model ;44 and 344 directl. Two locations which are geographicall separate and having ver di$erent temperatures can have same number o# degree das. 5ricing o# weather derivatives reGuires the #uture value o# local temperature hence we should have the abilit to predict regional weather conditions #or the coming months. ;ence an e$ective model o# variations o# a weather derivative contract over the course o# #uture months is essential #or pricing o# the contract. 7inancial traders develop a #orecast o# the economic conditions be#ore #ormulating a trading strateg. Similarl the people who trade in weather derivatives contracts reGuire #orecasts o# temperatures e8pected in the #uture ie metrological #orecasts. A variet o# models are made use o# b the weather #orecasting frms involving man parameters to predict the weather conditions. Short term Hs Iong term predictions. -egional Hs 6lobal predications. /ven though man models o# #orecasting the weather conditions e8ist but still accurate predictions is not possible and is #ull o# uncertaint. ;owever some short?term trends can be predicted. ;ence one can predict todaAs weather with more certaint as compared to tomorrowAs weather and tomorrowAs weather. Similarl tomorrowAs weather can be predicted with more certaint as compared to ne8t weekAs weather. Iong?term weather #orecasting reGuires thorough understanding o# past weather patterns and seasonal e$ects. The major controvers in the weather derivatives market is the choice o# pricing methodolog used which will help us ascertain the #air value o# the di$erent derivative contracts. There is no one such pricing methodolog which is widel used and accepted b everone.
%& '#ac()Sch!#es an" *eather "erivatives& 7isher %lack and ,ron Scholes developed option pricing model which is used to determine prices o# 3all and 5ut options and is used currentl also. Un#ortunatel! the %lack?Scholes model is based on certain assumptions that do not appl realisticall to weather derivatives. 0ne o# the main assumptions behind the model is that the underling o# the contract <in our case ;44 or 344= #ollows a random walk without mean reversion. 1n other words! their model predicts that the variabilit o# temperature increases with time! so temperature could wander o$ to an level whatsoever. 1n the fgure attached below! we can see di$erent simulated dail temperature values #or a three month period assuming that there is no mean reversion. The simulated temperatures di$er substantiall #rom e8pected temperatures! and we can see how the variabilit increases with time. 9e can see how these simulated temperatures are totall unrealistic! since towards the end o# the simulation! we have temperatures as high as 1*' and as low as ' degrees #or the same da o# the ear. The %lack?Scholes model is probabl inadeGuate #or weather derivatives #or the #ollowing reasons+ J 9eather does not KwalkL Guite like an asset price Krandom walkL! which can in principle wander o$ to Bero <think o# degrees .elvin= or infnit <hotter than the sun=. 1nstead! variables such as temperature tend to remain within relativel narrow bands! probabl because o# a mean?reverting tendenc! i.e. a tendenc to come back to their historical levels. J 9eather is not KrandomL Guite like an asset price random walk. %ecause o# its inherent nature! weather is appro8imatel predictable in the short run and appro8imatel random around historical averages in the long run. This means that short?dated weather derivatives ma behave #undamentall di$erent than their long?dated counterparts. J %lack?Scholes option pao$ is determined b the value o# the underling e8actl at the maturit o# the contract. 9eather derivatives usuall provide #or averaging over a period o# time! and are there#ore are more akin to KAsianL or average price options! i.e. have a non?%lack?Scholes pao$s. J ,an weather derivatives are also capped in pao$! unlike the standard %lack?Scholes option. J The underling variables <e.g. temperature or precipitation= are not tradable prices! and so pricing cannot be #ree o# econom risk aversion #actors! unlike the %lack?Scholes model. +& Simu#ati!ns ,ase" !n hist!rica# "ata !r -'URN ANA./SIS The Kburn analsisL approach is ver simple to implement and tries to answer the Guestion+ 9hat would have been the average pao$ o# the option in the past D earsM The main objection is that it does not incorporate temperature #orecasts in its pricing. Using the earl series o# historicall realiBed cumulative degree?das over the relevant instrument period we can determine the e8pected pao$ #or each ear. The #air price o# the option would be the average o# those historical pao$s. 0& M!nte Car#! 'ase" Simu#ati!ns K,onte 3arloL is a computer?based method o# generating random numbers which can be used to statisticall construct weather scenarios. Such ,onte 3arlo simulations provide a Ne8ible wa to price di$erent weather derivatives structures. Harious tpes o# averaging periods! such as those based on cumulating ;44s or 344s! can be specifed easil. Similarl! and as easil! a contractual cap placed on the price o# the derivative can be taken into account. ,onte 3arlo tpicall involves generating a large number o# simulated scenarios o# ;44s or 344s to determine possible pao$s #or the instrument. The #air price o# the instrument is then the average o# all simulated pao$s! appropriatel discounted to account #or the time value o# mone. 7or ,onte 3arlo based simulations! it is important to choose the right random process #or temperature. 1t is reasonabl clear that temperature is mean reverting! and there#ore an models that onl assume %lack? Scholes stle Krandom walkO behavior will be inadeGuate to model temperature. 1ndeed! measurement o# the reversion rate parameter in temperature data indicates that temperatures tend to revert to normal levels in & or ( das. Weather Derivatives in In"ia 9e have seen the how popular the weather derivatives have been U.S.A. there the major customers #or weather derivatives have been utilit companies but b 3hicago mercantile e8changeAs own admission the real potential will be tapped when #arming related activities start using the weather derivatives. 9eather derivatives have been launched in 1ndia as well. ;ere the major customers will be the #arming communit. 9e donAt see a huge potential #or the derivatives b utilit companies as 1ndia is a power shortage nation and we donAt see the chances o# e8cess power due to cooler summers or warmer winters. The reasons wh we think the weather derivatives will be success in 1ndia are+ P 7armers P Agriculture credit o$?take in ninth plan : -s. &!(1!79> crores <grew Q &'@ pa=R Target #or D plan : -s. 7!(S!)7' crores P 9'@ crop losses on account o# weather related risks P -ural /conom is highl weather dependent P 3ommodit Traders P 9eather related suppl bottlenecks make dr?land commodities ver volatile P 1ntrada volatilit o# 6uar! chill touches 1'?1)@ <dail trading at national e8changes touches -s.1''' crore dail= P Hegetable and #ruit ,andis highl dependent on temperature <4elhi ,andi trade alone touches -s.1''' crore annuall= P Trader income dependent on weather vagaries P 1ndustries like agro?input companies! #ood processing industr! companies! plantations! 7,36! %anks! 5ower sector etc P 2ot uncommon to fnd Agri?1nput companies! whose sale dips b over ('?*'@ due to Nuctuation in rain#all 0ne more thing that we need to change be#ore we implement weather derivatives in 1ndia is to have them #or rain #all changes as well The major U.S frms that trade in weather derivatives are utilit companies whose business is more a$ected b temperature as compared to rain#all patterns. 9hereas in 1ndia #arming communit is more dependant on rain#all patterns as compared to the temperature Nuctuations. Some compan reports supporting how weather impacts their business P /14 5arr sales! net down >S pc on monsoon #ailure. - The Hindu, Jan 17, 2003 P The 3ompanTs business is seasonal in nature and the per#ormance can be impacted b weather conditions ? Notes to !!ounts, "yngenta #$% &td' P ,onsanto 1ndia continued its strong proft growth on the back o# positive all? round business per#ormance aided b a good monsoon. - nnual (eport 2003-0), *onsanto &td P The delaed monsoon has hit the #ertiliBer stocks badl. - nalyst, Hindu Business &ine P +ver 1000 ,armers !ommit sui!ide in vidar-ha and Telangana in last t.o years / T+$ P An average drought costs upto -s * bn to the state e8cheGuer! 6ujarat earthGuake resulted in direct damages o# about -s.1)( billion :24,3 P Agricultural loss in man parts o# the countr is weather dependent P 9eather 4erivatives can fll in the gap P Ioss can be monitored real time P 3ost o# risk trans#er can be reduced through weather trading P 9eather <esp. rain#all= is the common commodit across diverse agri? products! industries P /8plains up?to large variation in prices #or commodities in the dr land P /ntities on both the long and short side 3ommoditC 1nde8CScrip /8tent o# linkage with economU Trading Turnover <in " Trillion=U 7ore8 4erivatives " 7.) tr o# #oreign trade Appro8. "1)''?1S'' trillion 1nterest rate derivatives " &'?&) trillion o# bond /Guit 4erivatives " 1'?1) trillion o# eGuit port#olio 1llegal %etting ??????????? " &.1) trillion <vis?V?vis! a total annual global savings o# "7.) tr= 9eather 4erivatives " 1)' bn <#or 1ndian econom= M P Iinkage o# the underling with econom is important P /nsures buers W Sellers P %ase liGuidit #urther deepens the market P 9eather impacts appro8. P 645 o# "1)' bn in 1ndia P 645 o# " &''?&)' bn in 1ndia W 3hina P 645 o# " *''?*)' bn in top > developing economies 4hat needs to be done to establish a mar&et for 0eather deri5ati5es in India 7irst thing that needs to be done is to create the much needed cashCspot market in 1ndia. Then we need to create an active #uturesCoptions trading market in 1ndia 4eepening the 5rimar market P Technolog development P -esolving the ke constraints 4eveloping the secondar market in tandem P Iaunching the 1ndices #or ke regions P Approaching the ke market segments P 3ommodit #unds! Agri?#unds! -ain#all speculators! 1nternational trading #unds P 5ush #or regulations on participation b %anks and ,71s P 5resence in both the 0T3 and e8change traded market P 4eveloping the ;brid market P Xuantos! Satellite 1mage : weather indices! 9eather?Area ield P 3at indices -eal Time 4ata availabilit #or an given long?lat 4ecision Support sstem<s= #or major customer segments 5lat#orm enabling Trading and cost reduction ,arketing 2etworkC-elat ionships To achieve these goals we need to cover important agricultural Bones real time! at a cost o# appro8. -s.)'' per sG.km or -s.) per hectare. This fgure has been worked out b weather risk management services. 9e also need to generate historical records #or an given longitude latitude positions. Statistical W 2eural 2etwork model models need to be developed and implemented Since weather derivatives are needed #or specifc areas we need to collect data #or each particular region. Since 1ndia is a monsoon dependant econom we there#ore need to stud water imbalances in each cit and region #or pricing the contracts 5rocessing o# data received #rom sensors and converting it according to specifc product 0n Site wireless Sensors 7armers 3onsole <61S= This illustration shows that impact o# rain is di$erent during di$erent phases o# vegetation. Since there will be contracts with di$erent maturit dates these #actors need to be kept in mind be#ore we price a derivative Another #actor that should be kept in mind is the modeling e8treme weather risk. 9e need to keep in mind these losses and thus prepare models to identi# and evaluate the risks caused on account o# e8treme weather conditions. This will be something in line with value at risk model and stress testing which is used #or the traditional derivative contracts So considering all this #actors and combining it with the satellite images! we will get something like this which we need to look at while pricing the derivatives. 9e will need the standard deviation and the mean o# the rain#all and temperature #or the area to be able to compute the price using 6aussian model #or pricing o# derivative contracts. He"ging Weather Ris( Using Weather Derivatives Lets consider there is a company which is facing revenue shortage due to abnormally warm winter. The company maintained enough reserves against normal variations in temperature but not for many continuous warm winters. One more warm winter can be ruinous for the company. He therefore decides to hedge the risk using weather derivatives. The company is located in a region where weather records are available and the nearest measurement site is 100km from the city.. The first thing that the company does is to derive the time series data of 5 winters measured in the city. !ased on the data" the following things become clear# the town from where the company operates is a cold place. $verage monthly temperatures in winter can fall to 1%&' ()5&*+" with fre,uent temperatures below -ero. There is almost perfect correspondence between average winter temperature and .ovember through /arch heating degree)days (correlation coefficient is 0.01+. There is no clear and convincing trend in these 5 winters" and Three warmer)than)average winters in a row have happened twice before in this last half century" but four in a row have not. $lso the forecast for the coming winters is that it will be warmer than normal. $fter obtaining the weather pattern we need to study the basis risk of the company between heating in the company2s distribution region and the weather measurements. 3e also need to find out the correlation between the temperature at the measuring site and the regional weather. 3e also need to find out what is the temperature which leads to an increase in temperature which results in an increased demand. 3e also need to ,uantify the natural weather e4posure of the company. 5nowing company6s weather e4posure will help know the risk associated with it. The chart above shows the gross revenue levels off in e4tremely cold weathers ( the upper curve+.however when there is e4treme demand the company runs out of fuel and has to buy more from the open market at higher cost. The net revenue can go negative( the lower curve+. 7t can be seen from the graph that the critical net revenue for the company is not -ero but million. The net revenues become negative when there are less than %850 heating degree days and when it is more than 9050 hdd6s .e4t we need is the probability distribution of the weather conditions. :sing the daily temperatures of last thirty years we simulate the forecasts. :sing the data available in the e4ample we see that 1%; of the times the H<<6= are less than the critical figure of %850 while >; times H<< are more than 9050" the two critical values. To hedge this risk we need to buy an out of money H<< call which will pay if winters are too cool and sell a near the money swap that will pay if the winter is warm. The company will have to pay some cash for the call and sacrifice some revenues in the swap deal but only if winter is cold and revenues are good. Taking the 0 yr average H<< we price the swap near 5185 H<<. To ensure million revenue we need to price each H<< at 10"000. The call should be bought below 9050 and should compensate both for revenue loss and the premium paid for shorting a swap. 3e can calculate the strike using some software. =uppose the strike works out at 5150 and the >0"000 for each H<<. The fair value works out at 85"000. !y multiplying the revenue at each degree day occurrence with the probability of that occurrence we get the probable revenue curve. This way they can ensure that their revenue never goes below the critical revenue. Thus by buying a swap and selling a call helps them to hedge the risk of revenue falling below the critical million mark no matter what the weather conditions are.