Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized

PolicyResearch

WORKING PAPERS InternationalTrade InternationalEconomicsDepartment The WorldBank January 1993 WPS 1077

Risk Management Prospects for Egyptian Cotton Panos Varangis Elton Thigpen and TakamasaAkiyama

The New Yorkfutures marketdoes not providean appropriate mechanismsfor hedgingthe pricerisk in Egyptiancotton under present proceduresfor determiningprices. Establishinga domesticspotmarket(whileprivatizingtheindustry),followedby a forwardmarket, may providethe best interim mechanism.

PlbicyRescatchWolkingPapendissminatethe&indingsofwo*inprogres andencuuragetheexchangeofideasa.nongBank staffand an othes increstedin dtclopmentissues.Thesepapa, distributed bythcResarch AdvisoryStaff carm y thenamesoftheauthors,refuect

only 'weirviews,and should bcuseand dted scoordingly Thefindings.inteqrprtations, and conclusionsgaretheauthors'own.Theyshould not be attuibutedto the Wodld Bank. its Board of Dis,m its managyeme, or any of its member countrieL

Policy Research

InternationalTrade WPS 1077

This paper-a product of the Intemational Trade Division, International Economics Department-is part of a larger effort in the department to explore possibilities for commodity risk management in developing countries. Copies of the paper are available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please contact Dawn Gustafson, room S7-047, extension 33714 (January 1993, 26 pages). Varangis, Thigpen, and Akiyama examine risk management options for Egyptian cottons, the export prices for which are volatile. They use regression analysis to establish whether Egyptian cotton's prices can be effectively hedged by using existing futures contracts on the New York Cotton Exchange. They find no relationship between the movements in prices of Egyptian long and extralong cottons and prices for the base quality of U.S. medium staple cotton traded on the New York futures market. (Probably because Egyptian cotton prices are government-detennined, U.S. medium staple cotton prices are influenced by nrice support policies unrelated to the longer staple markets, and the fiber of the cottons analyzed have different physical characterist!s.) So, the New York cottcn futures market's No. 2 contract is not an appropriate mechanism for hedging the price risk facing Egyptian cotton under present procedures for determining prices and probably not under market-determined prices. If the cotton market in Egypt is liberalized, cotton prices there may correlate more with prices elsewhere - especially for the longer staple cottons.

Varangis, Thigpen, and Akiyama extend their regression analysis to the prices of other medium staple cottons - Australian, Central Asian, Mexican, Pakistani, and Turkish - to determine how they behave relative to U.S. medium staple cotton prices. None of these prices had short-term movements closely related to U.S. cotton prices, indicating mainly the influence of domestic policies on the U.S. market. Again, the New York futures No. 2 contract does not provide a satisfactory hedge for these cottons. The cotton futures contract recently introduced in New York (world cotton contract) based on the Cotlook A Index - may prove useful for hedging the price risk for some cottons (especially Australian, Central Asian, and Pakistani) but apparently not Egyptian cotton. Varangis, Thigpen, and Akiyama recommend (together with privatizing the industry) establishing a domestic spot market to give transparency to the price-forming process. When the spot market is functioning well, establishing a forward market could provide a hedging instrument for Egyptian cotton.

The PolicyResearchWorkingPaperSeriesdisseminatesthefndings of workunderwayin theBank.Anobjectiveof theseries is to get these findingsout quickly, even if presentationsare less than fully polished.The findings,interpretations,and conclusionsin these papersdo not necessarilyrepresentofficialBankpolicy. Producedby the Policy ResearchDisseminationCenter

RISK MANAGEMENTPROSPECTSFOR EGYPTIANCOTTON By P. Varangis, E. Thigpen and T. Akiyamna InternationalTrade Division The World Bank

RISK MANAGEMENT PROSPECTS FOR EGYPTIAN COTTON

Table of Contents Introduction ..........................................

1

I. Characteristics of the Egyptian Cotton Market ..............................2 HI. Relationshis Between Ctton

......................................... 8

11. Explanation of the Statistical Results

.

............................. 13

IV Issues in Establishing an Egyptian Cotton Exchange .................18 V Condusions .........................................

24

References ..........................................

26

1

Introduction This paper examines risk managementpossibilitiesfor Egyptian cottons. Risk managementis important given that the volatility of export prices for Egyptian cotton is high. The coefficient of variation is between 18.6% and 23.4%, depending on the cotton variety, while the average for all primary commoditiesis between 17%and 19%.' Three questionsarise with respectto risk management: first, can Egyptiancottons be hedged using the existingcotton contract in New York? If the answer is negative, can Egyptian cottons be hedged using a futures contract representing other cottons such as Turkish, Pakistani or Central Asian cottons? Third, if price movements of Egyptian cotton are uncorrelatedwithprices of other cottons, what is the possibilityof establishinga cottonfutures contract(s) representingEgyptian cotton varieties? The paper investigatesthese questions.

The structureof the paper is as follows:the first sectiondescribesthe market for Egyptiancottons and for other extra-finecottons. It also describesthe characteristicsand structure of the cotton market in Egypt. The secondsectionpresentsan econometricanalysisof the relationshipbetweenthe movements of the pricesof Egyptianand other cottons. This sectionoffers answersto the first two questionsabove. Sectionthree discussesthe results of the aboveanalysisand assessesthe implicationsfor hedgingoptions. Sectionfour deals with the feasibilityof establishinga cottonexchangefor spot and futures transactions. It describesthe constraintsfacing the establishmentof a successfulfutures contract for Egyptian cotton, as well as presentingsome ideas of how a spot market could operate. Sectionfive concludes.

'The coefficientof variation is measured as the standarderror of the regression of the price on its time trend.

2 1. Characteristicsof the Egyptian Cctton Market

The Egyptiancotton market operated activelyunder free market conditionsprior to the socialist revolutionin the early 1950s. Duringthe pre-revolutionperiodan activephysical spot market for cotton was in operation in Alexandria. It was maintainedbriefly in the post-revolutionperiod but became inoperative due to heavy government interference. During the pre-revolutionperiod, Egypt was the world's principalproducerand exporter of extra-longand long staplecotton. It accountedfor over 70% of world productionand exports in these cottons.

In Egypt, cotton is typicallyproducedon small owner-operatedfarms. Cottongrowinghas been limited to 33% of total crop land and farmers have been required, by law, to plant a specified area to cotton. However, the cotton area has now declined to around 13.3% of the toWalcrop area. The Governmentdeterminesthe area to be planted, the cotton crop rotation, and the variety which farmers in a given locality may grow. It also buys seedcottonfrom farmers at procurementprices set annually. Cotton procurement prices are proposed by the Ministry of Agriculture and approved by the Higher Policy Committeewhich representsthe main ministries.

The Egyptian Holding Company for Cotton, formerly the Egyptian Cotton Authority, in the Ministry of Economyand Foreign Trade controlscotton marketing,ginning, and exports. The Cotton Authority exercises these functions through supervision of six export companies, five ginning companies, and a cotton press/balingplant, all of which are public sector organizations.After harvest, farmers deliver their seedcottonto village cooperativecollectioncenters. Seedcottonis transportedfrom there to a nearby gin. Farmers are paid on the basis of lint classificationafter deduction of ginning, transportation,and cooperativecosts. Until recently, the price received by farmers was equivalentto

3 about50% of the world price. Under the recent economicreforffiprogram,'cotton producerprices have been raised to 66% of world prices.

Cotton lint is classified by the Cotton Arbitration and Testing Generai Organizationagainst maintainedstandardsaccordingto variety, grade, and type. The qualityclassificationsystemfor Egyptian cotton containsas many as 16 grades, but currently the export pricing system includes prices for only seven grades of the Giza 45 variety and for only the five top grades of all other varieties.

After ginning, cotton lint is allocated to export companies for sale to foreign buyers and to government-ow ad spinning milis by the Ministryof Economy and Foreign Trade, in consultationwith the TextileIndustriesCorporation,now also a holdingcompany,and the Ministryof Industryand Mineral Wealth. Allocations are based on the size of the crop, domestic mill requirements, cotton imports, Egypt's need for foreignexchange, and internationalprices. Cottonexportprices are set by the Ministry of Economyand Trade and the Cotton Authority,based on an assessmentof world market conditions. When prices are announced,bids are invited from foreign buyers. Domestic spinning mills buy lint at highlysubsidizedprices, which are usuallyequivalentto or lower than prices receivedby farmers.

The Textile Industries Corporation (TIC), a public sector holding company, has a virtual monopolyon cotton yarn productionand controlsaround 90% of the weavingindustry. It has 30 large, mostly vertically integratedenterprisesunder its direction. The installed capacity in the spinning mills consistsof three millionring spindles(design capacityof 3 million tons/year) and 25,000 rotors (design capacityof 40,000 tons/year). The TIC has a monopolyon all cotton procurementfor domesticuse.

4 Under the ongoing economicreformprogram, the Governmenthas undertakensignificantsteps towards liberalizingthe cottonsector. The basicframeworkof the reforms includes: (a) unificationof the exchange rate at the commercialrate; (b) gradually increasing input prices to reach economic levels within a specified time period; (c) gradually lifting import bans and reducing the number of banned commodities;(d) revising the tariff code; (e) removalof informaltrade restrictions;and (f) assurancethat public and private enterprises face the same incentives.

Egypt's cottonproduction accountsfor only about 2% of the world total and its share of world output of extra-finecottons (extra-longand long staples) has declined to less than 30% during the last three seasons. Due to the desirable qualitiesof thesecottonsfor manufacturinghigh-valueproducts, they commandpremium prices in world markets relativeto other cottons. Moreover, prices for these cottons appear to change in responseto supply and demand for these specific qualitiesand to be little affected by changesin prices for medium staple cottons.

Egypthas had a long traditionof growingand exportinga major share (about40%) of the world's longestand finestcottons. Onlyafter the late 1970shas extra-finecottonproducedoutsidethe Nile River ValleychallengedEgyptiancottonsin a major way (Tables 1 and 2). Peru has producedextra-finecotton for many years, but since 1970output has fluctuatedwidely, while only occasionallyexceeding30,000 tons. Historically,India and the former SovietUnion (FSU) were major importersof extra-finecotton but since 1986/87, having developed their own production capabilities, they are exporters of these cottons. Chinahas increased its productionof extra-finecotton in recent years, from 15,000tons in 1986 to 60,000 tons in 1990. Some of this cotton is offeredfor export. Israel's productionof extra-finePima cotton fluctuatesin response to the availabilityof irrigation water and the rt, ative prices of Pima and upland cottons.Its output is virtually all exported. The most importantcompetitorsfor Egyptiancottons in export marketshave been SudaneseBarakat (until recently)and United States Pima (since 1989/90).

.

.... ..... .

5 Its characteristicsof fiber finenessand strength make extra-finecotton desirablefor spinningthe highest quality combed cotton yarns. Such yarns are used for producing fine apparel fabrics, laces, embroideryfabrics, knitted goods, householdfabrics, and strong industrial cloths. The unique qualities of extra long staple(ELS), and to a lesser extent long staple (LS) cottons,limit the competitionfaced by Egyptian cottons to a few select extra-fine fiber cottons that account for a small proportion of world cotton production(5.3% to 7.5% in recent years).

Table 1: PRODUCTIONOF EXTRA-FINECOITON 1986TO 1990 ('000 Tons)

1989

1990

25.0 18.5 23.0 40.5 72.8 191.2 390.1 10.8

35.1 31.2 32.6 49.1 150.7 213.1 264.0 11.0

60.0 15.5 26.8 14.2 78.9 194.0 318.0 14.1

82.5 265.2

80.6 226.2

80.8 204.2

82.7 209.3

1152.3

1101.9

1078.7

1071.6

1012.8

7.5

6.2

5.9

6.2

5.3

34.5

31.6

28.4

26.6

28.8

Country/Year

1986

1987

China Israel Peru Sudan United States India FSU Others

15.0 16.0 28.1 73.1 44.8 254.6 313.0 10.1

25.2 12.5 10.7 42.4 62.0 217.6 371.0 12.8

Egypt, ELS Egypt, LS

109.3 288.3

World Total Extra-Fine's Share of World Cotton (%) Egypt's Share of Extra-Fine Prod. (%)

1988 |

Source: CottonWorld Statistics,ICAC, April 1992

6 Syntheticfiber yarns are sutitablefor"someextra-finecotton uses and have made serious inroads into former all-cottonproducts.2 The extremely high prices for extra-finecotton during recent seasons and competitionfrom sy-.thetics,in thread and fabric manufacturinghave reduced import demand for cotton. Therefore, world exports have declined from their rather stable level of 373-387,000tons per year during the period from the early 1960sto the mid 1970sto an average of 275,000 tons during the 1987-90period (Table 2). Moreover, in recent seasons, a substantial shift has occurred in the export supply of extra-fine cotton. The declining exports of Egypt and Sudan have been replaced by other producers, particularlythe UnitedStates. Egypt's export supplyof extra-finecottonhas diminisheddue to decliningproduction and domesticmills taking an increasingshare of output.

Since the extra-fine cottons are, in a sense, a specia' y fiber, they are employed in manufacturing in a specialized sector of the general textile industry and many countries do not manufacturethese cottons. The numberof countriesthat importthese cottonsin significantquantitiesis even more limited because several consuming countries fill their own needs-such as India, the Commonwealthof IndependentStates, and the UnitedStates. The major importers of fine cottonsare WesternEurope and Japan.

2 Even in extra-finecotton's largestend-use-sewing thread-synthetic monofilamentis used as a core

of the thread to add strength. The synthetic's weaknessin this use of a low meltingpoint is overcome by wrappingthe thread's syntheticcore with cottonyarn to protect the syntheticcomponentfrom the high temperaturegeneratedin high-speedindustrialsewing operations.

7 Table 2: WORLD EXTRA-FINECOTTONEXPORTSBY ORIGIN 1987/88TO 1990/91 ('000 Tons)

Country

1988/87

1988/89

1989/90

Egypt United States Sudan FSU India Israel China Peru Others

87.7 51.6 63.1 47.5 NA. 12.5 21.8 2.5 11.3

60.0 57.7 52.5 56.5 NA. 18.0 13.0 11.1 13.8

43.0 98.4 32.9 31.6 25.5 30.5 8.2 10.1 7.5

18.0 90.4 27.7 7.8 3.6 14.2 10.0 20.7 11.7

Total

298.2

282.6

287.6

204.0

Notes: Egyptianexports are for ELS and LS cotton combined. The cotton season begins August 1 of the year designated. NA indicatesdata is not available. Source: CottonWorld Statistics, ICAC, April 1992.

1990/91

8 II. RelationshipsBetweenCotton Pricas We employedregressionanalysisto establishwhetherfluctuationsin Egyptiar.cottonprices (ELS and LS) can be hedged using the New York cotton futures market. With the regression analysiswe sought to establishwhether the price changes in Egyptian cotton could be explainedby changesin US cotton prices. We regressed the logarithmic difference of each of the Egyptian cotton prices on the logarithmicdifference of the US cotton price. We used the logarithmic difference because these price 3 For hedging purposes, the differencesin the price levels do not matter; what series are non-stationary.

matters is whetherprice movementsin the Egyptianand US cottons are closely related. The US cotton price analyzedis for grade 41, staple 34 (mediumstaple) cotton. We chose this quality because it is the base quality for the New York No. 2 cotton contract and, thus, the New York cotton futures' price has to follow this spot price closely. If prices for other cottonsmove closely with the US cottonprice, the New York cotton futures contract can be used for hedgingthe price risk of these other cottons.

The results of the regression analysis are presented in Table 3a. No relationshipwas found between the Egyptian ELS and LS cotton price movementsand movements of the US medium staple cotton price. The R-squaresare zero and the coefficientsare statisticallyinsignificant. Therefore, the N.Y. cotton futures market's No. 2 contractis an inappropriatemechanismfor hedgingthe price risk of Egyptiancotton. There are two plausibleexplanationsfor the lack of a significantrelationshipbetween the prices of these cottons. First, Egyptian cotton prices are determinedby governmentpolicies and, thus, their movementsare unrelatedto cottonpricesdeterminedin the US market.' Second, the technical

3 Non-stationarityimplies that

regressions in levels (or log levels) may be of a spurious nature (Phillips, 1986). This implies that neither the parameter estimatesnor their standard errors shouldbe tested. To avoid this problem, regressionsshould employdifferencedor log-differencednon-stationary data. 4See Monke (1981).

9 characteristicsof Egyptian ELS and LS cottons are greatly different from those of US medium staple cotton, and they do not substitutefor each other in the manufactureof cotton textiles. Therefore, their prices are formed under different supply/demandconditions.

Similarregression analysiswas performedon the movementsof prices of AmericanPima cotton and the prices of US medium staple cotton. Since American Pima cotton is substitutablefor some qualitiesof Egyptian cottonsand its export price is determinedin the internationalmarket, a significant relationshipbetween the prices for American Pima and US medium staple cotton would hold some promise for hedging Egyptian cotton on the N.Y. futures market once the Egyptian cotton market is privatized and the impact of government intervention on the prices of Egyptian cottons is absent. However, the regression analysis (Table 3a) does not reveal any relationshipbetween Pima and US mediumstaplecotton price movements. This result indicatesthat the presentcontractat the N.Y. futures market is not appropriatefor hedging Pima cotton, nor would it be useful for hedging Egyptian cotton price risk in a liberalizedmarket.

If price movementsare not close from monthto month, are there market forces that at least pull themtogether in the long run? This questioncan be analyzedusing tests for cointegration. Cointegration examineswhetherthere exists a long-run stationary relationshipbetween two or more variables. It can also be consideredas a test for market integration. That is, in Section I we saw that Egyptian cottons have different characteristicsthan other, mediumstaple, cottons. The questionis, are these differences enoughto implymarket segmentationin cotton? The results of the tests for cointegrationbetweenprices of Egyptiancottons,pima cotton, and US mediumstaple cottonare shown in Table 4. These tests reject the hypothesisof a long-run relationshipbetween the two Egyptian prices and the US medium staple cotton price. However, such a relationshipbetweenPima and US medium staple cotton prices cannot

10

be rejected, at least at a 90% level of significance. Hence, over the long run Pima and US cotton prices tend to movetogether. It appearsthat market forces, such as substitutabilityin production, tends to pull these prices together in the long run (betweenseasons), even if their prices diverge in the short run (within seasons). Such a relationshipdoes not hold between Egyptian cotton prices and US medium staple cotton prices. It is importantto note that for hedgingpurposes, the month-to-monthrelationship matters. Cointegrationtests are used to see whether there is a broader, long-run relationshipbetween prices. These latter tests give an indicationof the extent of integrationof the world cotton market.

We extendedthe regressionanalysisto see how other cottonprices benavevis-a-vis the US cotton price. We chose medium staple cottons of comparablequality from Pakistan, Turkey, Central Asia5 , Australiaand Mexico. The results showedthat none of the five prices investigatedmoves closely with the US grade 41, staple 34 cotton price, indicatingthe influenceof domestic policies and US weather conditions on the US market.' All R-squares are quite low. Hence, the New York futures contract providesan unsatisfactoryhedgefor all of these cottons. The strongestshort-runprice relationshipswere found between the Central Asian, Mexican, Pakistani, Australian and Turkish cottons. However, the cointegrationtests performedindicatedthe existenceof long-runrelationshipsbetweeneachof theseprices and the US cotton price (althoughit was not as strong for the Australiancotton as for the others).7 In summary, the cointegrationtest results show that all cotton prices, except the Egyptian, tend to move together in the long-run.

5 Until recently, the Cotton Outlook called this cotton "Russian".

MThe impactof the US Governmentinterventionis somewhatpredictableby market participants.The US Governmentaffects the market mainlythrough subsidiesto reduce export and domesticprices when they exceeda certainlevel, limits on acreageand the loanprograms, whileall these instrumentscan make US prices diverge from other cotton prices in the Europeanmarket, they do not impede the functioning of a futures contract. Although, they may limit the use of cotton futures contracts. 'Note that transitivity holds in cointegration. If X is cointegratedwith Y and Y with Z, then X is cointegratedwith Z.

11 The analysisdescribedabovesuggeststhatttheNew York cottonfutures7canbe usedto hedgeonly the medium stapleUS cottonvarieties. This can explainwhy the New York cottonfutures market is used primarily by US traders, textile mills, and cotton growers, and less frequently by foreign interests. In other commodities,such as coffee,cocoaand crudeoil, futures contractin New York are commonlyused by foreignproducers and consumers.

The recent introductionof a cotton futures contract in New York based on the Cotton Outlook "A" Index may prove to providean acceptablehedge for a numberof cottons. We employedthe same regression analysisas before to explore this issue. Table 3b shows that the Cotton Outlook "A" Index (CLAI) canprovide an adequatehedge for the Pakistani,Austialian and Central Asiancottons, but is less suitable for the Mexican and particularlythe Turkish cottons. As is the case with other cottons, CLAI is poorly correlated with either the ELS or the LS Egyptian cotton prices. Furthermore, it was found to be poorly correlated with Pima cotton prices. The CLAI movementsare also not closely related the movementsin US grade 41, staple34 spot prices. So the introductionofthe CLAIcottonfutures contract does not appear to be suitable for hedgingthe long fine cottons such as Egyptian and Pima varieties.

Finally, we tested the relationshipbetweenEgyptian cottons and the Turkish and Central Asian cottons. If a relationshipexists, the establishmentof a regional futures contract could be to the mutual benefit of the cotton sectors in these countries. However, the regression analysisindicatesthat there is no short-run (month-to-month)relationshipbetweenthese prices. Nor does the cointegrationtests show that there is a iong-run relationship.

12 Another way to look at the relationshipbetweenprices is to comput. the correlationcoefficients (CC) betweenpairs of prices.8 However, becausethe price levelsare non-stationaryit is not appropriate to compute the CCs on their levels, rather it should be done on their percent differences. Table 5 presents the computedCCs for the period 1985to 1991,using monthlydata.9 The CCs that were found highly statisticallysignificant are those between the Central Asian, Turkish, Australian, Mexican and Pakistanicottons. The US cotton (spot)relationshipwith the Australian,Pakistani,Mexicanand Central Asiancottonswas also significant. The weakestrelationshipwas foundbetweenthe Egyptiancottonsand Pima with every other cotton price in our sample. The relationshipbetweenthe two Egyptian cottons, ELS and LS, is significant. Theseresults are consistentwith those obtainedfrom the regressionanalysis.

%Regression analysislooks at the sameshort-runrelationshipas the CC. Cointegrationtests investigate long-run relationships. 9 The

statistical significance of the CC is computed by multiplying the CC by the number of observations. The computedstatisticis distributedat X2 (1). In our case, each CC should be multiplied by 72 and the critical value of the XI (1) is 3.84 at the 95% level of significance.

13 III. Expilanationof the StatisticalResults The independentmovementof prices for US mediumstapleand Egyptianextra-longstaplecottons is as expected.Those two types of cotton are not substitutablein their primary uses. Moreover, the price of each is influencedby independentdomestic cotton policies that address different market situations. Monthly price movementsare also affected by dissimilar seasonal marketingpatterns. During several recent seasons, the export supply of Egyptian ELS cotton was quite limited and its sales were typically allocated to traditionalusers at prices determinedby the Egyptian Government. The marketingseason was short--oftenno more than one or two months--andprices during the remainderof the season were of no informationalvalue, being set at levels near those at which earlier sales took place. By contrast, price movementsof US mediumstaple cotton appearedto be the result of changesin supplyand demand in the domestic and export markets during the entire year. The Egyptian ELS and LS cotton prices movedtogethermainly becausethe Governmentof Egypttended to set a fixed differentialbetweenthem each season.

The weaknessof the relationshipbetween prices for US Pima cotton and US medium staple cotton basicallyreflects their lack of substitutabilityin manufacturinguses. The area planted to Pima cotton does appear to be largely determined by the relative prices of Pima and medium staple cotton. However, this is an annualdecisionand has only a limitedinfluenceon monthlyprice movementswithin a season. Therefore, the number2 contracton the New York CottonFutures Market, based on medium staple US cotton, is not a satisfactoryprice hedgingmechanismfor either US Pima or for other cottons that competewith US Pima in internationalmarkets.

14 The divergenceof monthly cotton export price movementsfor four medium staple cottons from major exporting countriesrelative to US domestic medium staple cotton prices are related to divergent domesticcottonpolicies. The US market for raw cotton is largely insulated from foreign suppliesdue to import restraints which are relaxed only in periods of very tight supply. Cotton exports from Pakistan, Turkey, and Central Asia were closely controlledduring the early parts of the seasdns being analyzed to assure supplies to domesticmills. Variousgovernmentalagenciesimplementedquotas or variable tax rates on cotton for export.

The regressionresults indicateda relativelystrongrelationshipbetweenthe short-termmovements of export prices of medium staple cottons from Australia, Central Asia, Mexico, Pakistan, and Turkey. This is an expectedresult since these prices represent the offer rates of cottonsof similar quality in the world's largest importing region--WesternEurope-under very competitive conditions. An important implicationof this result is the possibilityof operating a successfulregional cotton futures market for hedgingpurposes. The necessaryvolumefor such a market could potentiallybe providedby production in Central Asia, Pakistan and Turkey.

Despite the low correlationbetween the short-run price movementsof US and other cottons, a strong long-run relationshipexists due to their close substitutabilityin the manufactureof cotton textile products and the competitiveness of cotton transactions in import markets. The slightly weaker relationshipbetweenthe monthlyprice movementsfor US and Australianmediumstaple cottonsover the long-runcouldbe related to the differingseasonalpatternsfor marketingcrops grown in the northernand southern hemispheres.

15 Table 3a: REGRESSIONRESULTS'°

Variables

Coeff.

t-stat

R2

D.W.

SER

PERIOD 1983M9 - 1991 M7 ELS, SPOT ELS, PIMA LS, SPOT LS, PIMA PIMA, SPOT ELS, LS

-0.00 0.06 -0.00 0.07 -0.01 0.88

-0.07 0.91 -0.02 1.05 -0.28 21.40

0.00 0.01 0.00 0.01 0.00 0.83

1.98 2.02 1.14 1.99 1.56 1.94

0.04 0.04 0.04 0.04 0.04 0.02

PERIOD 1981M9 - 1991M7 PAK, SPOT IZMIR, SPOT MEX, SPOT* AUS, SPOT RUS, SPOT*

0.25 0.12 0.16 0.15 0.18

4.77 2.12 2.81 2.91 3.08

0.16 0.04 0.13 0.10 0.13

1.2U 1.49 1.65 1.22 1.03

0.06 0.06 0.06 0.05 0.06

LS, IZMIR ELS, IZMIR ELS, RUS* LS, RUS*

0.09 0.09 -0.02 -0.02

1.44 1.45 -0.23 -0.23

0.02 0.02 0.00 0.00

1.93 1.94 2.03 1.99

0.04 0.04 0.04 0.04

PAK, IZMIR RUS, IZMIR* RUS, PAK* AUS, PAK AUS, IZMIR AUS, RUS* MEX, AUS*

0.53 0.43 0.78 0.68 0.39 0.81 0.68

6.45 4.05 14.05 12.74 4.06 14.21 8.89

0.26 0.21 0.75 0.73 0.21 0.77 0.53

1.37 1.17 1.77 2.18 1.34 1.77 1.82

0.05 0.06 0.03 0.03 0.05 0.03 0.04

Notes: ELS refers to Egypt's Menoufi/Giza70 extra long staple cotton; LS refers to Egypt's Dendera/Giza69/81 long staple cotton; SPOT refers to UnitedStates grade 41 staple 34 cotton which is the representativeprice in the New York exchange; PIMA refers to Americanpima G3 1-7/16" extra long staple cotton; PAK refers to the PakistaniSing/Punjabcotton; IZMIR refers to the Turkish Izmir ST 1 white 1-3/22RG cotton; MEX refers to the Mexicanmiddling1-3/32staple cotton, and AUSrefers to the Australianmiddlingstaple 1-3/32"cotton; RUS refers to Central Asian (formerly referred to as Russian)Vtoroi, medium staple cotton. All prices are in US cents per pound, CIF North Europe. *

Regressionswith Central Asian and Mexicancottons were for the period 1985 M9 to 1991 M7.

'the regression is of the form Y, = a. + a, X, + u,, where Y, is the log differenceof the first variablein the columnof variables, X is the log differenceof the secondvariable, a., a,, are coefficients and u, is the error term. The column coefficientrefers to the value of a, and the next column, t-stat, refers to t-statisticfor a,. D.W. stands for the Durbin-Watsonstatisticfor serial correlationand SER is the standarderrorof the regression.

16

Table 3b: REGRESSIONRESULTSFOR COTLOOK- A INDEX AND OTHER COTTON PRICESPERIOD 1981M9- 1991M7

Coeff.

t-Stat

R2

D.W.

ELS LS

A INDEX A INDEX

0.06 0.06

0.81 0.74

0.01 0.01

1.97 1.94

IZMIR PAK RUS ARTS MEX SPOT PIMA

A INDEX A INDEX A INDEX A INDEX A INDEX A INDEX A INDEX

0.60 1.07 1.00 0.86 0.63 0.86 0.02

6.25 18.50 19.27 14.05 7.74 5.20 0.21

0.25 0.75 0.84 0.74 0.46 0.19 0.01

1.76 2.02 1.48 2.07 2.05 2.23 1.56

See notes at the bottomof Table 3.

Table 4: COINTEGRATIONTEST RESULTS

DF

ADF

ELS, SPOT ELS, PIMA PIMA, SPOT LS, SPOT

-1.23 -0.99 -1.57 -0.31

-1.02 -1.10 -2.75* -0.27

PAK, SPOT IZMIR, SPOT PAK, IZMIR RUS, SPOT AUS, SPOT MEX, SPOT

-3.20** -3.51** -4.19** -3.00** -2.86* -2.74*

-3.00* -3.21** 4.68** -2.94* -2.71* -3.56**

ELS, IZMIR LS, IZMIR

-1.17 -0.40

-1.28 -0.44

Variables

#*

Significantat the 90% level. Significantat the 95% level.

17 Table 5: CORRELATIONMATRIX OF COTTONPRICE MOVEMENTS

dLELS dLELS dLLS dLPIM dLSPOT dLPAK dLIZM dLRUS dLAUS dLMEX dLAOUT

dLELS dLLS dLPIM dLSPOT dLPAK dLIZM dLRUS dLAUS dLMEX dLAOUT

dLELS dLLS dLPIM dLSPOT dLPAK dLIZM dLRUS dLAUS dLMEX dLAOUT

dLLS

dLPIM

dLSPOT

1.000 0.950 0.086 -0.014 -0.061 0.017 -0.027 -0.073 -0.003 -0.018

0.950 1.000 0.081 -0.007 -0.040 0.035 40.028 -0.062 -0.006 -0.012

0.086 0.081 1.000 -0.038 -0.120 -0.055 -0.079 0.017 -0.122 -0.079

-0.014 -0.007 -0.038 1.000 0.401 0.180 0.354 0.320 0.339 0.404

dLPAK

dLIZM

dLRUS

dLAUS

-0.061 -0.040 -0.120 0.401 1.000 0.539 0.864 0.844 0.777 0.892

0.017 0.035 -0.055 0.180 0.539 1.000 0.446 0.459 0.545 0.500

-0.027 -0.028 -0.079 0.354 0.864 0.446 1.000 0.854 0.716 0.918

-0.073 -0.062 0.017 0.320 0.844 0.459 0.854 1.000 0.730 0.860

dLMEX

dLAOUT

-0.002 -0.005 -0.122 0.339 0.777 0.545 0.716 0.730 1.000 0.681

-0.018 -0.012 -0.079 0.404 0.892 0.500 0.918 0.860 0.681 1.000

NOTES: The prefix dL signifies logarithmicdifference. ELS and LS are the two Egyptiancottons. PIM is the US pima cotton, SPOT is the US 41,34 mediumstaple cotton, PAK is the Pakistanicotton, IZM is the Turkish, RUS is the Central Asian (formerlyRussian",AUS is the Australian, MEX is the Mexicanand AOUT the Outlook A index. For a precise descriptionof each of the series see the notes of Table 3a. Correlation coefficients higher than 0.07, in absolute terms, are statistically significant at the 95% level of significance.

18 V. Issues in Establishingan EgyptianCoton Exchange Given the weak relationshipbetweenthe short-run price movementsof the Egyptian (ELS, LS) cottons and the US cottons, Egyptian cottons cannot be hedged using the New York cotton futures market. Thus, the questionarises whether there is room for the creationof an Egyptian cotton futures exchange? The Governmentof Egypt (GOE) is currently consideringre-openingthe Cotton Exchange in Alexandriawhich was closed in the early 1950s. This would be part of a program to liberalizeand privatizethe cotton industry. GOEhas plans for the Cotton Exchangeto offer servicesin futures as well as in spot market transactions.

The main benefit to Egypt of establishingsuch an exchange is that it would determineprices for Egypt's high quality cotton in a transparent and competitivemanner. At present, prices for Egypt's cotton are set by the Ministry of Economyand Trade and the Cotton Authoritybased on an assessment of factors including world market conditions,the size of the crop, domestic mill requirements,cotton imports, and Egypt's need for foreignexchange. Egypt's cottonexport prices have been set at very high levelsin recent years and, althoughshort-termdemandis inelastic,the high prices helped to dramatically reduce export volumes.

For a futures market to operate effectively,there are a number of conditionsthat must be met." 2 (b) clear and appropriate Probably the five most importantare: (a) a well establishedphysicalmarket;1

trading rules and the establishmentof a well-functioningclearing house; (c) freedom from government intervention; (d) adequate liquidity, i.e., large numbers of users and high volumes; (e) confidencein

"For a general and extensivediscussionon these conditionssee Leuthold (1992). '2 A well established physical market is one where prices are transparent and determined competitively.

19 standardizedgrading; and (f) support of the commoditybusiness community. Becauseof difficultiesin satisfyingthese conditions, futures markets in various commoditieshave failed.

The basic functionsof a futures exchangeis to develop, publish and enforce the rules of trading. The rules of trading and their enforcementaim at making the system trustworthy and thus encouraging the participationof traders.13 Any systematicattemptto influenceprices via controlson the sector will damage the trust traders place in the functioningof the futures market. An exampleis the intervention of the US Governmentin commodityexchangesduring the 1960s. Through price support and stocking policies, the US GovernmentcontrolledUS cotton prices and the traders' interest in the cotton futures market diminished,to the point where the New York cotton futures contract did not have the needed liquidityand becamealmost ineffective.

The issue of liquidityis very important. Liquidityassuresbuyers and sellers of futures contracts that whenthey wish to enter the market there will always be someonewilling to sell to themor buy from them. Note that for a futures market to be successful, it has to attract not only physicaltraders, as in the case of a spot market, but also speculators. The latter contribute significantlyto liquidity. It is illustrative,therefore, to look at figures for the New York cotton futures contract. In late April 1992, cotton futures contracts extended to March 1993. The total open interest outstandingon one day was about 34,000 contracts, equivalent to 770,000 tons of cotton. The daily volume of transactionswas roughly 7,000 contracts or 159,000 tons. Annual US production of cotton is about 3.5 million tons. Roughlythen, the open interest amountedto 22% of annual productionand the daily trading volume to 4.5% of annual production. In Egypt's case, if only the cotton that is exported is traded on the futures '3 Theclearinghouse has as its main function,the maintenanceof guarantydeposits, the maintenance of the original and variation margins, the recording and posting of transactions and also handling of physicaldeliveries.

20 exchange, roughly 50-55,000 tons annually, it is very doubtfulthat there would be sufficient liquidity. If the numbers for the United States applied to Egypt, only 1.5 to 2.5 thousand tons would be traded. If, however, all cotton produced is included, then daily trades would increase to 14-15 thousand tons. Again it is doubtfulwhether this volume is large enoughto support a futures market.

The issue of liquidity should also be examinedin conjunctionwith the varieties and grades of Egyptian cotton. Egyptian cotton is very well classified. The exported ELS cotton consists of four varieties, each of which has five grades (Giza 45 has seven grades). The exported LS consist of two varieties with five grades each. In terms of magnitudes,ELS production was about 83,000 toils and exports 27,000 tons in 1990. The figures for LS are 209,000 tons produced and 7.000 tons exported during 1990. Usually, the futures contract is based on a specificgrade within a variety with discounts and premiumsapplied to the qualitiesdelivered above and below the average. If two contracts were to be introduced, one based on the average variety and grade for ELS and the other on the LS, liquidity could be an even greater problem. However, given the results in Table 7, ELS and LS prices move closelytogether with a basis risk of about 17%, which is withinan acceptablerange. So for the sake of liquidity,one contractbased on the most prominentlytradedvariety and grade of the ELS couldrepresent both the ELS and LS varieties and deliverablegrades, with, of course, the appropriatediscounts and premium." Still, liquidity could be a significantconstraint--particularlyif only exports are included. However, it wouldalso be an importantconstraintevenif the whole cottonproduction(ELS and LS) was representedby this contract.

'4 Basisrisk signifiesthe difference in the movementsbetweentwo prices. Accordingto commodity brokers in New York, an acceptablebasis risk is below 20%.

21 Sincethe futures market is considerablymore complicatedfor traders than the present syst..m and it is questionablewhether there would be sufficient volume necessary for the effective operation of a futures market in Egypt, the CottonExchangeshouldstart with only a spot market. There are two types of spot marketsthat couldbe consideredfor Egypt's cotton; a physicalauction market and an electronic tender market. A physical auction market is one where commoditiesare brought to the market and auctionedoff. Examplesof this are the tea auction markets in Calcutta, Chittagong, and Mombasaand the coffee auction market in Nairobi. At present, there are no physical auction markets for cotton.

One of the main reasons for the non-existenceof physicalauction markets for cotton is that the same commoditytrading can be accomplishedmuch more efficientlyby a tender system using telex, fax or other electronicmeansof communication.An importantadvantageof the tender system is that neither the commodity(samples,in the case of cotton)nor the people involvedin the trade have to be physically at the same place where the sale occurs. Egypt and many other countries utilize tenders to purchase cotton for import. Tanzania's cotton is tendered using modern communicationfacilities. Tanzania's CottonBoard sends messagesto potentialbuyers providingdetailed informationabout the cotton to be sold. Bid openings and sales awards are public events. Unannouncedreserve prices are utilized to prevent sales at prices regardedas unacceptablylow. More efficientversionsof the auctionprocess have been used since 1975for marketing cotton produced in the US Southwest region (Texas plains and Oklahoma)through an electroniccommunicationnetworkmanagedby Telcot. Continuedinnovationsin communicationshave enabled improvementsin local coverageand the range of services provided.

It would be feasible for Egyptian cotton to be traded by tender within an anticipatedprivate marketingsystem. Cotton's durability,storabilityand establishedsystem for quality determinationmake it possible to trade cotton without the physical presence of buyers at the trading site. However, the

confirmationof the qualityof the deliveredproductmust be reinforcedby tradingrules that permit by agreedexperts. arbitrationof oualitydiasputes

in Egypt,GOEand For a cottontendertradingsystemto operateefficientlyand intemnationally the cottonindustrywouldhaveto makeseveralchanges,including: (a)

Introducemarketincentivesfor the operationof parastatalsand permitthe involvement of privatesectorcompaniesin both exportingand domestictrading. The privatization of the six exportingparastatalswhiledesirableis not essentialfor the tender trading into the systemto operate. More importantis the introductionof marketincentive-s cottontradingsystem.

(b)

Sellthroughthe systemfor domesticdemandas well as for export. The benefitwould be the increasedvolume,the transparencyof price discovery,and the integrationof domesticand exportprices.

(c)

If imposed,make exporttaxesexplicit,insteadof settingfixedprices withthe tax (or subsidy)beingthe residual. This, again,is necessaryto renderthe systemtransparent for everyoneinvolved.

privatebodyor by a cottonexporters' Sucha tradingsystemcouldbe operatedby an independent, associationonce the cottontradehas been liberalized.

23 After a physical spot market for cotton has been established and has earned the trust of the businesscommunity,a next step could be the establishmentof a forward market.5 A forward market functionslike a spot market, i.e., using the proposed auctionform, with the only difference being that the physicaldelivery of cotton will be sometimein the future. The establishmentof a forward market couldprovide a useful instrument for hedgingEgyptian cottons.

'5A forward market can function as a futures market in terms of hedging. Producers, by selling forward can 'lock" a price, the same way as if they sold futures. The two notable differencesbetween futures and forward markets are: first, forward markets usually require physical delivery. Second, forward markets do not require daily margins (marked-to-market),thus the performance risk can be significant.

24 VI. Conclusions This paper has shown that short-termmovementsin Egyptiancotton prices, extra-long stapleand long staple, are not correlated with movementsof cotton prices in the New York futures market. Therefore, Egyptiancottonscannotbe hedgedusing tiie New York cotton futures contract. Similar tests on AmericanPimaprices (an extra-longstaplecottonthat substitutesfor some Egyptiancottons)and US medium staple cotton prices showedno relationshipbetweentheir price movements. It was also found that no relationshipexists between movements in Egyptian cotton prices and prices of cottons from Turkey, Pakistan,Central Asia, Mexico,and Australia. This ledus to formulatetwo explanations. First, Egyptiancottonspossesssufficientlydifferentcharacteristicsfrom medium-staplecottonssuch as the ones tested that their price movementsare not correlated. Second, the prices of Egyptian cottonshave been so heavily influencedby governmentalinterventionthat their movements were divorced from market fundamentals. The government'srecent reformprogrammay increasethe correlationbetweenEgyptian and other cotton prices to some extent. This may be more the case for the LS Egyptian cotton than the ELS cotton, given that the LS cottonhas more substitutesin the internationalmarket.

Whenexaminingthe correlationsbetweencottonprice movements,we found thatprice movements betweenmedium-staplecottonsfrom variousorigins were correlatedamongthemselvesbut not withprice movementsin the New York No. 2 cotton futures contract. In effect, therefore, none of the other medium-staplecottons tested can be hedgedusing the New York cotton futures contract. However, the medium-staplecotton prices were correlated with the Cotton Outlook "A" index. So, the recent introductionof a New York futures contract based on the Cotton Outlook "A" index may provide medium-staplecottons with a useful hedginginstrument--butnot the Egyptian cottons. Egyptian cotton price movementswere not correlatedwith movementsin the Cotton Outlook "A" Index.

25 In examiningwhether a futures exchangecan be establishedin Egypt for Egyptian cottons, the study found the followingobstacles: (a) lack of a well-estab!ishedphysical market; (b) the existenceof heavy governmentinterventionin domesticcommercializationand exports;and (c) in all likelihood,lack of adequateliquidityfor establishmentof a futures contract. For these reasons it may not be wise to try to establisha futures contractfor Egyptian cotton. Rather, it is recommendedthat Egypt should try to establisha spot market in order to bring transparency to price-setting. A necessarycondition for such a market to function soundly is for the private sector to play a significantrole. If a spot market is ,-stablished,after some time the next step can be to set up a forward market that couldprovide a hedging instrumentfor Egyptiancottons.

26 REFERENCES Coleman,J. and M.E. Thigpen, "An EconometricModel of the World Cottonand Non-CellulosicFibers Markets", World Bank Staff CommodityWorkingPaper, No. 24, 1991. InternationalCotton Advisory Committee,Cotton: World Statistics,various issues. Leuthold,R., "AssessingAccessto Futuresand OptionsMarkets", paperpresentedin "Risk Management in LiberalizingEconomies: Issues of Access to Futures and OptionsMarkets", proceedingsof a workshop, organized by the World Bank, Paris, France, June 3-5, 1992. Monke, E.A., "Interactionof Demand, Supply and GovernmentPolicy: A Case Study of Egyptian Cotton", Working Paper No. 2, Universityof Arizona, July 1981. Phillips,P.C.B., "UnderstandingSpuriousRegressionin Econometrics",Journal of Econometrics,1986, 33, pp. 311 40. Tligpen, M.E., "InternationalCottonMarketProspects",World BankStaff CommodityWorkingPaper, No. 2, June 1978.

Policy Research Working Paper Series

Tltle

Contact for paper

Author

Date

WPS1060Economiesof the FormerSoviet Union:An Input-OutputApproach to the 1987 NationalAccounts

DmitriSteinberg

December1992

E. Zamora 33706

WPS1061 DynamicResponseto Foreign Transfersand Terms-of-Trade Shocksin Open Economies

KlausSchmidt-Hebbel Luis Serven

December1992

A. Maranon 31450

WPS1062 EconomicDevelopment, EnvironmentalRegulation,and the InternationalMigrationof Toxic IndustrialPollution,1960-88

RobertE. B. Lucas DavidWheeler HemamalaHettige

December1992

WDR 31393

WPS1063 Mongolia:Privatizationand System CevdetDenizer Transformationin An Isolated Alan Gelb Economy

December1992

A. Oropesa 39075

WPS1064 MoreEvidenceon Income Distributionand Growth

GeorgeR. G. Clarke

December1992

R. Martin 39065

WPS1065 StrengtheningUganda'sPolicy Environmentfor Investingin UniversityDevelopment

ThomasOwenEisemon John Sheehan GeorgeEyoku FranklinVanBuer DelaneWelsch Louisa Masutti Nat CoHletta Lee Roberts

January 1993

C. Cristobal 33640

January1993

A. Estache 81442

WPS1066 PollutionControlin a Decentralized Antonio Estache Economy:WhichLevelof KangbinZheng GovernmentShouldSubsidizeWhat in Brazil

WPS1067 Returnsto Investmentin Education: George Psacharopoulos January 1993 A GlobalUpdate

G. Psacharopoulos 39243

WPS1068 EnterpriseReformin EasternEurope Swedervan Wijnbergen

January 1993

B. Pacheco 37033

WPS1069 Why Is There ProportionatelyMore EstelleJames Enrollmentin PrivateSchoolsin Some Countries?

January 1993

D. Evans 37496

WPS1070 EconomicFundamentalsof Road Pricing:A DiagrammaticAnalysis

TimothyHau

December1992

WPS1071CongestionChargingMechanisms: An Evaluationof CurrentPractice

TimothyHau

December1992

Policy Research Working Paper Series

THle

Contact for paper

Author

Date

BrookeR. Johnson Janie Benson Janet Bradley AuroraRabagoOrdonez CatiaZambrano LeonardOkoko Leticia VazquezChavez PaulinaQuiroz KhamaRogo

January 1993

0. Nadora 31091

WPS1073 FiscalAspectsof Developing PeterJ. Montiel CountryDebt Problemsand Debt-Service ReductionOperations:A Conceptual Framework

January 1993

R. Vo 31047

WPS1074 How Movingto WorldPricesAffects the Termsof Tradein 15 Countries of the FormerSovietUnion

January 1993

N. Artis 38010

WPS1075Many Pathsto SkilledEmployment: AdrianZiderman A ReverseTracerStudyof Eight RobinHorn Occupationsin Colombia

January 1993

PHREE 33680

WPS1076 A PresumptivePigouvianTax on GunnarS. Eskeland Gasoline:Analysisof An Air Pollution ControlProgramfor MexicoCity

January 1993

C. Jones 37699

WPS1077Risk ManagementProspectsfor EgyptianCotton

January 1993

D. Gustafson 33714

WPS1072 Costs of AlternativeTreatmentsfor IncompleteAbortion

DavidG. Tarr

PanosVarangis EltonThigpen TakamasaAkiyama