New Developmentsin Financial Management

Dutch Rate Preferred Stock

Michael J. Alderson, Keith C. Brown, and Scott L. Lummer

Michael J. Alderson is Assistant Professor, Departmentof Finance, Texas A&M University;Keith C. Brown is Assistant Professor, GraduateSchool of Business, Universityof Texas at Austin; and Scott L. Lummeris Senior Consultant,Manufacturers Hanover Trust Co., New Yorkand Assistant Professor, Departmentof Finance, Texas A&M University.

I. Introduction In a search for attractive -terminvestment reduce investmentrisk while generatingimpressive yields, many corporationshave become involved in returns. cash managementprograms that are designedto cap- An alternativeto hedged dividendcapture became italize on the corporatetax exclusionfor dividendin- availablein 1982, with the introductionof adjustable come. These "dividendcapture" programs allow the ratepreferred stock (ARPS). In principle,this instru- participatingfirm to earnafter-tax returns that far ex- mentoffered greater price stabilitythan fixed ratepre- ceed the yields of traditionaldebt-based investments. ferredstock, since the dividendwas adjustedperiodi- Unfortunately,an unhedgedposition in preferredor cally for changes in interestrates. Wingeret al. [6] commonequity is also riskierthan the usual money analyzedthe returnperformance of adjustablerate pre- marketalternatives. Recent research has examined ferredstock duringthe firsttwo yearsof its existence. methodsfor reducingthe risk of dividendcapture pro- Theirresults showed that adjustablerate issues have grams. Examplesof these strategiesinclude writing yielded slightlymore thanmoney marketinstruments coveredcall options on a marketindex, and hedging on an after-taxbasis. The incrementalreturns were individualstock holdings with positions in call op- small, however,relative to the considerablevolatility tions. Brown and Lummer[1, 2] and Zivney and Al- in holdingperiod returns experienced by each issue. derson[7] have foundthat these methodssubstantially The figuresappear to go a long way towardexplaining why adjustablerate preferredstock has fallen out of favor as a cash managementtool. The authorswould like to thank Robert A. Taggart, John D. Finnerty, A new has been the intro- and Bernard J. Winger for helpful comments and suggestions. All prominent development remainingerrors are our own. duction of dutch auction rate preferred stock

68 ALDERSON,BROWN, AND LUMMER/DUTCHAUCTION RATEPREFERRED STOCK 69

(DARPS). This instrumentwas specificallydesigned of funds than would have been providedwith debt to circumventmany of the flaws in earlieradjustable issues. Approximately$8 to $10 billion of adjustable rateequities. By providingthat the dividendyield be ratepreferred was issued betweenMay 1982 and De- reset more frequentlythan an ARPS issue, througha cember1984. In 1985, however,less than$1 billionof dutch auctionprocess, this securityenables the pur- ARPSwere brought to market.The spectaculardecline chaserto realizethe corporatedividend exclusion in a in the popularityof this instrumentis traceableto sev- nearriskless manner. eral factors.First, adjustablerate preferredstock was Thepurpose of this paperis to analyzethe character- adverselyaffected by the Deficit RecoveryTax Act of istics and investmentperformance of dutch auction 1984. Thatact increasedthe risk exposureof all divi- ratepreferred stock. The followingsection reviews the dend capture programsby increasing the required factorsthat contributedto the disappointingperform- holdingperiod for the dividendexclusion from 16 to ance of the original adjustablerate preferredissues. 46 days. The measurealso eliminatedthe so-called Section III explains the fundamental aspects of "75%rule" for mutualfunds. Under that provision, DARPS issues and the dutch auction process. The registeredinvestment companies whose dividendin- fourthsection analyzes data from a sampleof DARPS come exceeded 75% of their total revenue were al- auctionoutcomes, and comparesthe resultingyields lowed to designate100% of theirtotal payout as eligi- on these securitieswith those of moretraditional mon- ble for the 85% dividendsreceived deduction.That ey marketinvestments. The resultsshow a relationship enabledARPS mutualfunds to pass a limitedamount amongyields that,to ourknowledge, is unprecedented of ordinaryinterest income through to theirsharehold- in the cash managementliterature. Specifically, yields ers on a tax preferredbasis. Eliminationof the "75% on DARPS are found to be set in such a way as to rule" decreased the attractivenessof those mutual accommodatethe marginaltax ratesof boththe issuing funds, andthat of theirunderlying securities in a con- and purchasingcorporations. The paper concludes comitantmanner. with a discussionof the likely adjustmentin DARPS Manyproblems with the ARPS issues were caused dividendsyields underthe recenttax code changes. by the basic structureof the securityitself. Specifical- ly, the severemismatch between the 13 week II. The Decline of the ARPS Market reset and the 46 day requiredholding period for the Adjustablerate preferredstock was first issued in 85% exclusion led to considerableinterest rate risk mid-1982by ChaseManhattan and Manu- exposure.This was becausecorporate investors had an facturersHanover Trust Company. The typical issue incentiveto hold the stockonly long enoughto qualify was structuredto maintaina pre-specifiedspread be- for the dividendexclusion. For manyinvestors, there- tween its dividendyield and the highest yielding of fore, the desiredholding period for an ARPSissue was threedifferent treasury securities. Dividend yields on 46 days. If interestrates changed dramatically,the adjustablerate preferred stock were reset on a quarterly price of ARPS duringthe period betweenresets was basis. affected.As such, a corporateinvestor that planned on Corporatepurchasers of ARPS issues were entitled sellingbetween reset days faced a greatdeal of uncer- to excludefrom taxable income 85%of the taintywith respectto total return. received.2 Competitive market pressures tended to The popularityof ARPS issues may have also suf- drivethe pre-taxyield on these securitiesbelow thatof feredfrom deficiencies in the formulaused to resetthe comparabledebt instruments.Hence, issuers in low dividend.For most ARPS issues, the reset rate was tax bracketswere able to realizea lower after-taxcost tied to the maximumrate amongshort-term, interme- diate-term,and long-termTreasury securities. Given the term of 'In the work to follow, the term DARPS will be used exclusively to structure interestrates that has appliedin describe the general class of auction rate preferred stock issues. It recentyears, the resetwas effectivelytied solely to the should be mentioned, however, that DARPS can also be found under long-termrate. In contrast,rates on debt severaltrade names. For instance, ShearsonLehman underwrites Mon- analagous ey MarketPreferred (MMP) while Salomon Brothers uses the Dutch instruments(floating rate notes) are typicallytied to a Auction Rate Transferable(DART) acronym. short-termbenchmark, such as LIBOR. Considering the enormouspopularity of the variable-ratedebt in- 2Recentchanges in the tax code have reduced the corporatedividend it is thatthe consid- exclusion struments, likely investingpublic from 85% to 80%. The impact that this modificationhas on ers this equity-basedcash managementprograms will be explored in a subse- type of linkagepreferable to the use of a long- quent section. term rate. That being the case, if short-termrates 70 FINANCIALMANAGEMENT/SUMMER 1987

shiftedwhile long-termrates remainedconstant, the Exhibit 1. Dutch Auction Bid Rates for Initial and price of a share of adjustablerate preferredstock PotentialStockholders of DARPS changed.Hence, the reset regime of ARPS provided Initial Holding Purchase Purchase a less than perfect hedge against interest rate Bidder (# shares) Bid Rate movements. A 70 70 4.6% Declines in the creditquality of the issuerwill also B 60 60 4.9% cause a preferredstock issue to becomeless valuable. C 40 40 4.9% In this the most of the D 30 30 5.5% regard,perhaps damagingaspect E 50 4.6% structureof adjustablerate preferredstock was its in- F 30 4.8% abilityto hedge creditrisk. Since the spreadbetween G 40 4.9% the dividendyield and the Treasuryrate was fixed, H 30 5.0% ARPS holdersfaced the same degree of exposureto 200 350 creditquality as investorsin fixed-ratepreferred stock. The creditrisk factorbecame particularlyimportant, given thatthe creditrating of many ARPS issuersde- clined during 1984 and 1985.3 terest rate risk from the security. Perhapsthe most importantfeature of the dutchauction process, howev- III. Dutch Auction Rate Preferred Stock er, is that it eliminates the risk of declining credit In an attemptto addressthe observedshortcomings quality. If the default risk on a share of DARPS is of ARPS, several corporationsbegan to experiment perceivedby the marketto have increased,the yield with alternativeforms of floating rate equity. Dutch derivedat the next reset auctionwill rise to reflectthe auction rate preferredstock is the most successful heightenedrisk exposure,but the shareprice will re- productof thatexperimentation process. Since the first main at par value. offering in 1984, over $10.0 billion of the security DARPSdividends are reset every seven weeks when have been broughtto market,with about$5.0 billion all currentowners and potential buyers of outstanding issuedin 1986 alone. The primarydifferences between sharessubmit price bids at a centralauction location. DARPS issues and their ARPS predecessorscenter The followingexample demonstrates the mechanicsof aroundthe frequency and method of resetting the the dutch auction market for DARPS issues.4 In this amountof the dividendpayout. In contrastto adjust- example, the dividendyield is to be renegotiatedfor able ratepreferred issues, the dividendyield on dutch 200 sharesof a given stock. Exhibit 1 lists a series of auctionpreferred stock is recastevery seven weeks in a currentand prospectiveparticipants, along with the biddingsession involving both currentand potential rate bids which they have submitted.Present stock- purchasingcompanies. These featuresreduce the risk holdersA, B, C, and D have submittedbids ranging to the purchaserin a numberof ways. from4.6 to 5.5%. Prospectivebuyers E, F, G, andH First, the increasedfrequency in dividend adjust- haveentered bids between4.6 and5.0%. Since all the mentreduces the impactof movementsin the general currentshareholders have enteredan explicit reserva- level of interestrates. Even more significantis the tion yield, all 200 of theirshares are potentially avail- importanceof the alignmentbetween the lengthof the able for reallocation.If any currentstockholders had reset interval and the minimum holding period re- desiredto maintaintheir investmentregardless of the quiredto qualify for the tax exclusion on dividends prevailing yield, the supply of auctionableshares received.Given that the two areclosely matched,cor- would be reducedaccordingly, and the new 49-day poratepurchasers have little incentiveto divest their rate would be set by the remainingparticipants. DARPS issues prematurely.In conjunctionwith the Given thatbids have been receivedfor 350 shares, propertiesof the dutchauction which allow for divesti- tureat this attribute eliminatesin- par value, virtually 4Inthe usualenglish auctionprocess, bid prices are startedat a relatively low level and successively raised until an equilibriumis reached. Con- versely, under the dutch auction format, bids are initiated above the 3The financial difficulties of some ARPS issuers severely affected the eventual selling price and lowered until the marketclears. (For a more marketvalue of all ARPS issues in mid-1984. Problemsat Continental completecomparison of the two mechanisms,see Cassady[3].) DARPS Illinois and other institutions caused their respective adjustablerate issues are set in a dutch auction even though bids are initiatedat rates issues to lose considerable value. Those price drops reverberated below the eventual equilibrium. This is because yields and price are throughoutthe market, impairing working capital invested in ARPS inversely related, making the proceduretantamount to inception above across the board. the eventual equilibrium. ALDERSON,BROWN, AND LUMMER/DUTCHAUCTION RATEPREFERRED STOCK 71

Exhibit 2. Rankingof DutchAuction Bids andSelec- prespecified"collar" on the dividendyield. If investor tion of ShareAllocation demandis insufficientrelative to supply,the dividend Purchase Cumulative Purchase Cumulative yield is typicallyreset to 110%of the prevailingyield Rate Allocation Total Bidder Bid Total on double-Acommercial paper. A situationin which 4.6% 70 70 A 70 70 all purchaserselected to retaintheir shares would con- 4.6% 50 120 E 50 120 verselycause demandto exceed supply, and the divi- 4.8% 30 150 F 30 120 dendyield wouldusually be resetto 58%of the current 4.9% 30 150 B 60 210 double-Acommercial issues also 4.9% 20 200 C 40 250 paperyield.5 Many 4.9% 0 G 40 290 protectthe purchaserfrom the effects of financialdis- 5.0% 0 H 30 320 tressby insuringthe issue with a letterof credit.That 5.5% 0 D 30 350 insurancefeature serves to protect investorsfrom a 200 350 deteriorationin the creditposition of the issuer, a fac- torwhich strongly influenced the disappointingmarket performanceof ARPS. IV. Empirical Analysis the dividendyield will be set withinthe rangeof bids indicated.Exhibit 2 demonstratesthe allocation Priorresearch has analyzedthe risk-returncharac- proce- teristics of several cash dure. Bid rates are in order, and equity-based management arranged ascending dividend rolls sharesare then allocatedto the bidders,starting with strategies, including preferred [5], dividend [1, 2, and 7], and adjustable the lowest yield bid and progressingupward. Within hedged capture the context of this the lowest bid is 4.6%. ratepreferred stock [6]. In this section, we extendthis example, list a of Consequently,70 of the 200 availableshares are allo- by examining sample auction-generatedyields catedto A. The continueswith 50 on DARPS.The resultssuggest that the pricingof this assignmentprocess new reflects the tax shares allocatedto E, 30 to F, andso on, untilthe securityclosely respective posi- being tions of the and Relative supplyof availableshares is exhausted.The bid rate issuing biddingparticipants. associatedwith the last shares to be allocatedis the to commercialpaper, DARPS yields are set in such a as to allocatethe benefitsof the exclu- rate."This rate, which is the highestof the way corporate "winning sion on dividendincome both the and successfullow bids, defines the dividendyield for all among issuing firms. The statisticsconfirm that the shares.In the presentexample, the winningrate is purchasing summary 4.9%. the instrumentis most advantageousto sellers in the Exhibit2 also shows how sharesare allocated when lowest marginalbracket, and to buyers taxed at the rate. demand at the winning rate exceeds the marginal highestmarginal The in this consistedof 201 sharesavailable. According to the rulesof the auction, sampleemployed study dutchauctions that were held over the first six months currentstockholders receive priorityover prospective of 1986.6For each of those the onesbidding the samerate, with the allocationmade in , winningrate, the commercial on the auction proportionto the respective purchasebids. For in- prevailing paperyield date, and the bill most corre- stance, 140 shares(60 + 40 + 40) are requestedat a Treasury yields closely to the seven week were col- rateof 4.9%, while only 50 (200- 150) are available. sponding holdingperiod lected. In the was of 60 As such, currentinvestors B and C receive 30 and 20 total, sample comprised sepa- rate issues from domestic For most shares, respectively,in proportionto their requested . four reset auctionswere availablefor (60:40), while bidder G re- firms, separate purchases prospective each issue containedin the ceives An interestingaspect of the dutchauc- sample. nothing. Exhibit3 a of the tion is thatthe relativemagnitude of the bids of A, E, provides summary comparative and after-tax for the dutchauction and F determinesonly whetheror not they get their pre-tax yields pre- desiredshares. Their rate bids could have been much ferredstock, commercialpaper, and Treasurybill in- vestments.Panel A containsthe calculationsfrom the lower, but their assigned securitieswould still have yielded4.9%. This illustratesthat the bid of the mar- ginalinvestor ultimately determines the yield on all the 5The figure 58% is derived as (1 -0.46)/(1 -0.069), which (prior to shares. 1987) was the proportionalpre-tax rate on DARPS that generatedthe same after-taxreturn to purchasersas commercial paper. The participantsin the dutch auctionare protected againstsevere imbalancesin supplyand demandby a 6Dataused in this study are available from the authorsupon request. 72 FINANCIALMANAGEMENT/SUMMER 1987

standpointof the issuer.The averageTreasury bill rate Exhibit 3. Summaryof ComparativePre-Tax and is the meancorresponding yield for the T-billmaturing After-TaxYields closest to the date of the next dutchauction, for each PANEL A Issuer respective issue. The average DARPS yield is the Average mean rate across 201 auctionevents. Since DARPS winning Commercial preferreddividends are not tax deductible,the after-tax Average Paper cost of DARPS to the issuer is equal to the pre-tax Marginal Average Average Commercial Spread yield. The averagecommercial paper yield is the mean Tax TreasuryBill DARPS Paper Yield (Basis Rate Rate (%) Yield (%) (%) Points) of the prevailing60-day commercialpaper rate corre- 0 7.273 202 spondingto the respectiveauction events, while the 6.532 5.251 after-tax is the returnof 7.273% 27.80% 4.717 5.251 5.251 0 yield averagepre-tax 31.17% 4.496 5.251 5.006 (24) timesthe quantityone minusthe marginaltax rate.The 34.00% 4.311 5.251 4.800 (45) differencebetween the averageDARPS yield and the 46.00% 3.527 5.251 3.927 (132) average commercial paper yield is the average PANEL B Purchaser DARPS-commercialpaper spread. It representsthe Average after-tax if of DARPS advantage(or disadvantage negative) Commercial auction-ratepreferred to the issuing firm.7 Average Paper PanelB detailsthe pre-taxand after-taxfigures for Marginal Average Average Commercial Spread The DARPS is Tax TreasuryBill DARPS Paper Yield (Basis purchasingcorporations. average yield Rate Rate (%) Yield (%) (%) Points) calculatedunder the recentlyenacted revision of the tax whichreduces the taxrate 0 6.532 5.251 7.273 (202) code, marginalcorporate 27.80% 4.717 5.032 5.251 22 to 34%, and decreasesthe exclusionfor dividendsre- 31.17% 4.496 5.006 5.006 0 ceived to 80%. An alternateyield is also calculated 34.00% 4.311 4.894 4.800 (9) underthe tax law prevailingat the time of the auctions, 46.00% 3.527 4.889 3.927 (96) which is based on a 46% marginalcorporate income All of the yields listed above are reportedon a 365-day, simple interest tax rateand an 85%exclusion for dividendsreceived. basis. The differencebetween the after-taxaverage DARPS yield andthe after-taxaverage commercial paper yield is the average DARPS-commercialpaper spread. It the averageafter-tax cost of floatingcommercial paper representsthe after-taxadvantage (or disadvantage, was 3.927%. This figureis 132 basis pointsbelow the whennegative) of dutchauction rate preferred relative after-taxcost of DARPS, which does not providethe to commercialpaper for the cash manager. interestexpense deductionattributable to commercial The averagepre-tax yields on DARPSand commer- paper. However, to the corporatepurchaser facing a cial paper were 5.251% and 7.273%, respectively. 46% rate, the effective tax rate on DARPS is only Thus,to an issuerfacing a low marginaltax rate,dutch 6.9%, owing to the 85% dividend exclusion. This auctionrate preferredstock is clearly the favorable causes the net returnto the corporatepurchaser of alternative.Conversely, the 202 basis point differen- dutchauction preferred shares to exceed the after-tax tial makesit disadvantageousto purchasersin low tax yields on commercialpaper by just under 100 basis brackets,since they will be unableto utilize the full points.Consequently, DARPS is advantageousto cor- potentialof the tax exclusion for dividendincome. porateinvestors with high tax rates,but unfavorable to Forissuing companies in the 46%marginal bracket, fully taxable issuers.8 One interestingaspect of the empiricalanalysis is the break-even tax ratesof issuersand inves- 7Commercialpaper and DARPS yields are quotedon a 360-day, simple marginal interestbasis while Treasurybill yields are quoted in 365-day, simple tors - that is, the tax rates at which the respective interestterms. All yields reportedin Exhibit 3 have been adjustedto a issuers and investors are indifferentbetween method using 365-day, simple intereststandard. The fact that this calculation commercial and DARPS. The after-taxcost of ignores compoundingmay affect the comparisonof yields between the paper commercialpaper, T-Bill and DARPS series (with respective average commercialpaper is equalto the cost of DARPSfor an maturitiesof 60, 51, and49 days). However, the impactof this omission on the tabulatedresults is negligible. For instance, the respective aver- age yields for the three securities calculated on an annualized, com- 8Althoughdutch auction rate preferredstock is more costly to fully pounded basis (see Fielitz [4]) are 0.07497, 0.06719, and 0.05372. taxablecorporations in termsof yield, it may neverthelessbe the appro- Consequently,the spreadbetween DARPS and commercialpaper only priatesecurity to issue when bond covenants prohibitthe sale of addi- increases from 202 to 212 basis points. tional debt securities. ALDERSON,BROWN, AND LUMMER/DUTCHAUCTION RATEPREFERRED STOCK 73

issuer with a marginalcorporate tax rate of 28%. is the directresult of the dutchauction process, which After-taxyields to commercialpaper and DARPS are allows the preferredstock's price to be reset at par equal for purchasersin a 31% marginaltax bracket. valueevery seven weeks. Whatuncertainty remains is Consequently,DARPS has on averagebeen advanta- an elementof liquidityrisk, inasmuchas there is, at geous to issuers who faced marginaltax rates below present,no formalsecondary market for DARPS be- 28%, and to purchaserswith marginaltax rates of tween auctiondates. Even this need not be a major above 31%.9 concern,though, since the seven week resetinterval is Althoughall of the auctionsdescribed above were virtuallyidentical to the holdingperiod required of any conductedprior to the 1986 tax legislation, we also equity-basedcash managementstrategy. computedthe after-taxyields underthe terms of the relevantrevisions to the InternalRevenue Code. As V. Conclusion they apply to corporateequity investments, these This paperhas examinedthe structuralcharacteris- changesreduce the maximumcorporate rate from 46 to tics andreturn behavior of dutchauction rate preferred 34%,and lower the exclusionfor dividendsreceived to stock.The resultsof the analysisshow thatthis securi- 80%.Recalculation of the after-taxspreads shows that ty providesa superiorafter-tax return relative to com- dutchauction preferred stock is only marginallybene- mercialpaper, with the same minimalamount of risk. ficial to the high-taxpurchaser, providing an addition- An analysisof the dataunder the termsof the Internal al returnof nine basis points, as shown in Exhibit3. RevenueCode of 1986 showsthat this relationshipcan We would expect the applicableterms of the recent be expectedto hold in the future. tax legislation to affect the yield spread between The empiricalresults of this study suggest that the DARPSand commercialpaper. Specifically, it seems tax benefits of the corporatedividend exclusion are likely that the yield on DARPS will rise relative to sharedbetween the purchaserand the issuer. It there- commercialpaper rates. That is becausethe tax advan- fore appearsthat dutchauction rate preferredstock is tageof a dutchauction preferred issue to a fully taxable an extremelyvaluable cash managementvehicle for corporateinvestor will be reducedfrom 39.1 (46% - fullytaxable companies to purchasefrom zero tax issu- 6.9%) cents per dollar of income, to 27.2 (34% - ers. Non-taxpayingcorporations would find it advan- 6.8%) cents. In termsof realizedyields, the total tax tageousto purchasecommercial paper from fully tax- advantageof DARPS relativeto commercialpaper is ableissuers. The recent surge in the amountof DARPS the sum of (i) the incrementalafter-tax return to fully being issued would indicate that those benefits are taxablecorporate investors, and (ii) the reducedcost of indeedrecognized by the marketplace. borrowingto the non-taxpayingissuer. FromExhibit 3, those gains are 202 and 96 basis points, respective- References for a total tax of 298 basis That ly, advantage points. 1. K. Brown and S. Lummer,"The Cash Management total Implica- woulddrop to about212 basispoints as a resultof tions of a Hedged Dividend Capture Strategy,"Financial the recentchange in the tax code. Assumingthat this Management(Winter 1984), pp. 7-17. advantagewill continueto be sharedin an approximate 2. "A Reexaminationof the CoveredCall OptionStrat- 2:1 ratio(0.02022:0.00962), corporationsshould still egy for CorporateCash Management,"Financial Manage- be able to issue DARPS at a cost of 140 basis points ment (Summer 1986), pp. 13-17. below prevailingcommercial paper yields. 3. R. Cassady,Auctions andAuctioneering, Berkeley, Univer- Finally, it must be noted thatthe precedingdiscus- sity of CaliforniaPress, 1967. 4. B. the Bond Yields for T- sion has concentratedexclusively on comparative Fielitz, "Calculating Equivalent DARPS returnsand the issue of investment Bills," Journal of Portfolio Management(Spring 1983), pp. ignored 58-60. risk. The for that over- simple explanation apparent 5. M. Joehnk, O. Bowlin, and J. "PreferredDividend is thatnone of the 201 auctionsthat were exam- Petty, sight Rolls: A Viable Strategyfor CorporateMoney Managers?", ined resulted in the loss of any invested capital or Financial Management (Summer 1980), pp. 78-87. dividendincome. This impressivedegree of certainty 6. B. Winger, C. Chen, J. Martin, J. Petty, and S. Hayden, "AdjustableRate PreferredStock," Financial Management (Spring 1986), pp. 48-57. x the breakeventax rates,these can be ob- 9Letting represent figures 7. T. and M. Dividend tainedby solving the equations(0.05251) = (0.07273)(1-x) and Zivney Alderson, "Hedged Capture (0.05251)[1-(1-0.85)x] = (0.07273)(1-x) for the issuerand pur- with Stock Index Options," Financial Management (Sum- chaser,respectively. mer 1986), pp. 5-12.