An Examination of Mispricing, Returns, and Uncertainty for Initial Public Offerings Author(s): Robert E. Miller and Frank K. Reilly Source: Financial Management, Vol. 16, No. 2 (Summer, 1987), pp. 33-38 Published by: Wiley on behalf of the Financial Management Association International Stable URL: http://www.jstor.org/stable/3666001 . Accessed: 14/05/2014 06:01 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

.

Wiley and Financial Management Association International are collaborating with JSTOR to digitize, preserve and extend access to Financial Management.

http://www.jstor.org

This content downloaded from 210.212.58.34 on Wed, 14 May 2014 06:01:35 AM All use subject to JSTOR Terms and Conditions

An

Examination

of

Mispricing, Returns,

and

Uncertaintyfor Offerings

Initial

Public

Robert E. Miller and Frank K. Reilly

The authors are respectivelyAssistant Professor of Business Administrationand BernardJ. Hank Professor of Business Administrationat the Universityof Notre Dame.

I. Introduction it was generallyagreedthatthese were attributable to the underpricing of these new issues by the underwriters. Manyof the studiesalso addressedthe implication of these results for the efficient markethypothesis (EMH).The generalconclusionhas beenthat,consistentwith the EMH, pricesadjustrapidlyto the underpricing,andinvestorswho purchasethe new issues in the after-market(i.e., a week or a month after the initial offering)do not experienceexcess returns. A more recent line of research(Beatty and Ritter [1], Ritter[19], Rock [20]) goes beyondthe empirical evidenceof underpricingand sets forththeoreticalargumentsthatexplainthe underpricingin termsof informationasymmetryamongtraders.The investor(or ex anteaboutthe trader)in theIPOeitheris "informed" after-marketequilibriumprice, or is "uninformed" aboutthe price. If an IPO is underpriced,informed traderswill enterordersforthe issue, causingthe issue to most likely be oversubscribed,andtherebyrequiring an allocationor rationingof the issue. Uninformed

An understandingof the marketfor initial public offerings(IPOs)is importantnotonly forinvestorsand underwriters,but also for financialmanagers.Small, non-publicfirms may need this marketin the future, for example. In addition,financialmanagersof curthismarketsince rentlypublicfirmsshouldunderstand there are numerousfirms going private and subsequentlygoing publicagain.Also, firmsaremorewilling to spin-offdivisionsto theircurrentstockholdersor allow managersto put togethera leveragedbuy-out thatmay eventuallygo public. Earlystudiesof IPOs[2, 3, 8, 9, 10, 11, 12, 15, 16, 17, 18, 22] weremainlyconcernedwithexaminingthe profitpotentialforinvestorsin theseissues. Theresults consistentlyindicatedthat, on average,IPOs offered significantpositiveexcess returnsin the shortrun,and The authorsacknowledgethe financial supportfrom the Jesse H. Jones Data Bank at the University of Notre Dame. We also appreciatethe commentsby participantsin the ResearchSeminarat NotreDame, anda special thanks to Robert Taggart for his comments and suggestions. 33

This content downloaded from 210.212.58.34 on Wed, 14 May 2014 06:01:35 AM All use subject to JSTOR Terms and Conditions

FINANCIALMANAGEMENT/SUMMER 1987

34

Exhibit 1. Issue Characteristics Mean

Shares Offered Offer Price FirmSize ($1000's)

1,480,307 $11.24 $67,159

Standard Deviation

Skewness

Minimum

Maximum

Median

2,024,170 $5.01 $104,439

8.28 0.34 4.63

240,340 $1.00 $3,400

30,252,144 $28.00 $1,159,773

1,000,000 $11.00 $32,101

traderswill thus receive some rationedamountin responseto their-orders. On the other hand, if an issue is overpriced,informedtradersstay out of the offering, leaving only uninformedtradersto absorb the overpricing.The of IPOsis thenconsideredto be a averageunderpricing to uninformed traderfor continuing the compensation to participatein the IPOmarket.Most importantfrom the standpointof ourstudy,thistheorypredictsthat,in an expectedvalue sense, underpricingwill be greatest for those issues thatare subjectto the greatestex ante uncertainty. It is the purposeof this study to re-examinethe speedof marketadjustmentto mispricing,andfurther to uncertainty. exploretherelationshipof underpricing Ourdataset allows us not only to studythese characteristicsin moredetail,butalso to relatetwo additional measuresof uncertaintyto underpricing.A description of the data set is presentednext.

II. The Current Study Therewere 1,106 IPOssold during1982-1983 that were not regulationA offerings.' For comparability, we consideredonly purecommonstockofferingsand stockswithaninitialofferingpriceof $1 ormore.With these constraints,there were 739 stocks. Finally, all stockshadto have bid andask pricesandvolumedata for each of the firstfive days, andfor the twenty-first day afteroffering.These priceseitherwere contained in Standard and Poor's Daily Stock Price Record or

were availabledirectlyfrom NASDAQ. Given these requirements,the final sample included510 stocks,2 which had the characteristicsshown in Exhibit 1. Clearlythe sampleincludesa diverse set of firms. 'RegulationA offerings do not exceed $1,500,000 in aggregatein any one year. They are exempt from detailed registrationrequirements. 2A comparisonof 196 of the issues not included in the study due to missing data with the 510 issues in the study revealed the following: therewas no significantdifferencein the numberof sharesoffered;there was a significantly lower offering price for those omitted ($8.66 vs. $11.24); and there was no significant difference in firm size. Furthermore, wherepartialdatawas available, no differenceswere found in the excess returnor the volume tradedmeasures.

All reportedreturnsarenet returnsequalto the percent pricechangefor the IPO minusthe percentprice change in the NASDAQ IndustrialIndex for OTC stocks duringthe same time period. First day IPO returnsare calculatedfrom the offeringprice to first day ending bid price, with subsequentdaily returns frombid-to-bidprices, unless otherwisenoted.

Ill. Results A. Analysis of Daily Returns Exhibit2 presentsthe returnresultsfor the firstfive daysof tradingfortheentiregroup.As canbe seen, the marketadjustmentto mispricingappearsto be accomplishedwithinthe firstday of trading.The averagenet returnof 9.87%is consistentwith the findingsof several previous studies that calculatedreturnsto one week or morefromthe offering.The fact thatnone of thedailyreturnsfordaystwo throughfive weresignificantimpliesthatexcessreturnsfoundin thoseprevious studieswereactuallyavailableonlyduringthefirstday of trading. Thesecondpanelof Exhibit2 containsreturnresults from the first five tradingdays for those issues that wereunderpriced (i.e., thoseexperiencingpositivereturnson thefirstday), whilethelastpanelgives similar informationforthe overpricedissues (thoseexperiencing negativereturns).The datarevealthatthe adjustmentto bothtypesof mispricingtakesplaceduringthe firsttradingday, with no significantreturnsoccurring on any of the subsequentdays. In Rock's [20] model of underpricing,it was asinsumedthat, for a cost, investorscould "purchase" formationaboutthe equilibriumpriceof an issue, and thusbecomean informedtrader.If this is so, then an informedtraderwould average15.39%, since such a traderwouldonly investin underpricedissues. On the otherhand, the uninformedtraderwould invest in all issues, averaginga 9.87% return.The differenceof 5.52% becomes an upperbound on the cost that an uninformedinvestorshouldbe willingto incurfor perfect information,i.e., to become an informedtrader. Exhibit3 containsthe excess returnsfromboththe offeringandthe end of the first day of tradingto four

This content downloaded from 210.212.58.34 on Wed, 14 May 2014 06:01:35 AM All use subject to JSTOR Terms and Conditions

MILLERAND REILLY/INITIAL PUBLICOFFERINGS

35

Exhibit 2. Daily Net Returnsfor Total Sample(510 Stocks) Day 1 2 3 4 5

Percent Average Net Return

Standard Deviation

Skewness

Minimum Return

Maximum Return

Median Return

9.87* 0.40 -0.18 -0.24 0.17

18.37 5.71 3.63 3.62 3.42

2.54 3.14 0.57 -0.02 0.16

-21.07 - 13.92 -11.61 - 18.66 - 13.49

124.94 58.40 19.26 21.20 15.18

2.78 -0.08 -0.17 -0.08 0.06

Daily Net Returnsfor SampleStocksthatExperiencedPositiveReturnson Day One (356 Stocks) Day 1 2 3 4 5

Percent Average Net Return

Standard Deviation

Skewness

15.39* 0.74 -0.02 -0.19 0.32

19.45 6.48 3.84 3.91 3.47

2.34 3.01 0.72 0.14 0.56

Minimum Return -

0.00 13.92 11.61 18.66 12.95

Maximum Return

Median Return

124.94 58.40 19.26 21.20 15.18

8.58 -0.06 -0.06 -0.22 0.10

DailyNet Returnsfor SampleStocksthatExperiencedNegativeReturnson Day One (154 Stocks) Day 1 2 3 4 5

Percent Average Net Return - 2.88* -0.39 -0.53 -0.34 -0.18

Standard Deviation 3.22 3.21 3.06 2.88 3.26

Skewness -2.20 -0.77 -0.37 -1.03 -0.98

Minimum Return

Maximum Return

Median Return

-21.07 - 11.74 - 10.88 - 12.09 -13.49

-0.01 7.53 11.31 8.95 9.62

- 1.77 -0.15 -0.26 0.07 0.00

at a 0.01 level. *Significant

weeks after the offering. This analysis indicatesthe profitpotentialfor investorswho are not able to get stockatthe offering,butacquireit in the after-market.3 The average net returnin the after-marketwas not statisticallysignificantat the 0.01 level. Theseresults areconsistentwith priorfindingsthatindicateit is not possible for investorsin the after-marketto benefit from the initial underpricing. When the after-marketreturnswithin each subgroupare examined,thereis a statisticallysignificant negative returnfor those issues that had a negative returnon day one. This suggests the possibility of abnormalreturnsby selling shortthose stocksthatdeclined at the end of day one and buyingthemback at the end of day 21 because the data reveal a 3.20% excess returnfromsuchtrades.However,severalIPO underwriters indicatedthatshortsales duringthe first in theafter-market werecalculated 3Returns usingthefirstdayaskprice andthe twenty-first day bid price.

Exhibit 3. Net ReturnsfromOfferingto FourWeeks After Offering(21 Days) for AlternativeGroups PercentAverage Net Return*

Standard Median Deviation Return

Total Sample 27.21 12.95t (-0.47) 6.19 Positive 28.32 15.59 20.17t (1.03) - 3.72t (- 3.94t) 14.15 -5.07 Negative arenetreturnsfromtheendof thefirstdayto *Figuresin parentheses fourweeksafteroffering(20 days). tSignificantat a 0.01 level.

severaldays of tradingwould not be permittedby the underwritingsyndicates.Thus, excess returnsin the after-market arenot availableto investorsfor the IPOs thatfell duringtheirfirst day of trading. B. The Relationship Between Uncertainty and Underpricing should BeattyandRitter[1] arguethatunderpricing be relatedto the ex ante uncertaintysurroundingthe

This content downloaded from 210.212.58.34 on Wed, 14 May 2014 06:01:35 AM All use subject to JSTOR Terms and Conditions

36

FINANCIALMANAGEMENT/SUMMER 1987

Exhibit 4. Daily Volume(00's) as a Percentof InitialOfferingfor Alternative Groups Total Sample 1,480,307

Positive Sample 1,518,540

Negative Sample 1,391,926

Size of Offering: Day

Shares

(%)

Shares

(%)

Shares

(%)

1 2 3 4 5 21

3266 1469 801 624 528 199

(22.1) (9.9) (5.4) (4.2) (3.6) (1.3)

3511 1686* 901* 694* 581" 217

(23.1) (11.1) (5.9) (4.6) (3.8) (1.4)

2700 970 569 463 400 159

(19.4) (7.0) (4.1) (3.3) (2.9) (1.1)

*Significantdifference between positive and negative samples' daily volume at a 0.01 level.

value of an issue. They tested for this relationship usingthe inverseof grossproceedsfromthe issue as a proxyfor ex ante uncertainty.Using the same uncertaintymeasure,we likewise founda significantpositive correlationof 0.12 in our samplebetweenex ante uncertaintyand the degree of underpricing(as measuredby the first-dayreturn).Ritter[19] also usedthe standarddeviationof returnsfor the first twentydays of tradingas an ex post measureof uncertainty.Using theex post standarddeviationof dailyreturnsfor days two throughfive as a riskmeasure,we founda significantcorrelationcoefficientof 0.32 betweenthis measure of risk and underpricing. Ourresultsthus supportthe propositionthatthe underpricedofferingsaresubjectto the mostuncertainty. A t-test on the differencebetweenthe standarddeviations of returnsfor days two throughfive for the versusthe overpricedsubsamplesyielded underpriced furthersupportfor this proposition,since there was significantlyhigher uncertaintyfor the underpriced group. To examine the risk-underpricing relationshipfor theunderpriced andoverpricedsubsamples,we calculatedcorrelationsbetweenfirst-dayreturnsandeachof the two risk measuresfor each groupseparately.For the underpricedissues, a correlationcoefficient of 0.15 forthe ex anteriskproxy,and0.30 fortheexpost risk proxy, were significantat the 0.01 level. The overpricedgrouphada correlationof -0.15 with the ex anteriskproxy,anda correlationof - 0.13 withthe ex post risk proxy. Both of these correlationswere insignificantat the 0.05 level. In additionto the use of the inverseof gross proceeds andthe standarddeviationof returnsas proxies foruncertainty,two additionalvariablesthatshouldbe related to uncertainty,tradingvolume and bid-ask

spread, were examined. The results are reported below. Analysis of Daily Volume Thepatternof daily volumeof tradingactivityis presentedin Exhibit4. The averagefirst day volumewas 326,600 sharesfor the entiresample.Thisrepresented22.1%of the average offeringissue of 1,480,307 shares.Sucha proportion for one day representsenormoustradingturnover comparedto normalsecondarymarketswherethetrading turnoveraverages about 30-40% a year.4 The averagevolumewas almosthalvedthe secondday, and by the twenty-firstday was downto 19,900 shares,or 1.3%of the offeringissue. A similarpatternheld for the various sub-groups.Overall, the volume results indicatedstrongtradingvolume on day one when all the significantpricechangesoccurred,which implies that volume and price movements(both underpriced and overpriced)were correlated. The level of tradingmay signifythe extentto which investorsdisagreeaboutthe value of a security.If the issuesaresubjectto thegreatestuncertainunderpriced this would ty, implythattheyshouldalsoexhibitgreater tradingvolume. We did a t-test of the daily mean differencesin tradingvolumebetweenthe underpriced andoverpricedgroups. The resultsin Exhibit4 indicate a significantlyhigher(0.01 level) level of trading for the underpricedgroup duringdays two through five. Even on day one andtwenty-one,the difference was significantat a 0.05 level. Hence, the largertradgroupprovidesfurther ing volumefor the underpriced evidenceof greateruncertaintyfor that group. Bid-Ask Spread The bid-askspreadis known 4Thisrepresentsthe average tradingturnoveron the NYSE duringthe last ten years as reportedin the 1985NYSEFact Book (New York, New York Stock Exchange, 1985).

This content downloaded from 210.212.58.34 on Wed, 14 May 2014 06:01:35 AM All use subject to JSTOR Terms and Conditions

37

PUBLICOFFERINGS MILLERAND REILLY/INITIAL

to be a functionof inventoryriskandadverseinformationrisk(see Stoll [21] andGlostenandMilgrom[6]). Inventoryriskdependsuponthepricevolatilityandthe numberof tradesfor each issue in inventory(see Ho andStoll [7]). Adverseinformationriskaffectsthebidask spreadsince market-makersare obliged to trade with investorswho might possess superiorinformation. GlostenandMilgromshowedthatlargerbid-ask spreadsexist when informedtradersare present(i.e., increasesthe spreadto compensate the market-maker for the greaterrisk of dealingwith informedtraders). Thisimpliesthatthe initialspreadshouldbe greaterfor becauseinformedtradersare IPOsthatareunderpriced involvedin these issues, in contrastto overpricedissues wherethere are only uninformedtraders. To capturepricevolatilitywe usedthe absolutevalue of dailyreturns.5Daily volumeactedas a proxyfor the frequencyof trades.6Stoll [21] observedthatthe discretenessof price quotes tends to overstate the quotedspread,andthatthebiasdecreaseswithincreasing price. To mitigate the effect of discrete price quotes,we includedthe daily bid pricefor each issue. We then estimatedthe following log-linearequation: + E, InSP = a0 + a-lInP + a2*IlnV + a3"lnR where SP = % bid-askspread; P = bid price; V = volume of trading; R = absolutevalue of return;and , = errorterm. Sincethereis not an explicitmeasureof adverseinformationriskin the model, the averageeffect of adverse informationrisk is capturedin the interceptterm.The resultsof theregressionforeachof thefirstfive trading days is reportedin Exhibit5. Each variablein the regressionwas significantfor eachof the firstfive daysof trading.An analysisof the underpricedandoverpricedsubsamplesusing dummy variablesrevealedthatthe sameestimatedcoefficients were applicablefor both groups,with one exception. Onday one, the underpricedstockshada significantly

Exhibit 5. Daily RegressionResults for Inventory RiskComponentof Bid-AskSpreadfor TotalSample Day

Intercept

1 2 3 4 5

4.54* 4.11* 4.33* 4.22* 4.34*

Price

-0.52* -0.49* -0.54* -0.55* -0.59*

Volume

Risk

-0.21* -0.19* -0.20* -0.19* -0.18*

0.08* 0.09* 0.10* 0.07* 0.07*

Daily RegressionResultsfor InventoryRisk Component of Bid-Ask Spreadfor UnderpricedIssues Day

Intercept

Price

Volume

Risk

1 2 3 4 5

4.97* 4.00* 4.69* 4.44* 4.63*

-0.44* -0.49* -0.51* -0.52* -0.59*

-0.26* -0.18* -0.24* -0.21* -0.21*

0.08* 0.10* 0.12* 0.06 0.09*

Daily RegressionResultsfor InventoryRisk Component of Bid-Ask Spreadfor OverpricedIssues Day 1 2 3 4 5

Intercept 3.90* 4.51* 3.95* 4.14* 3.92*

Price -0.70* -0.54* -0.59* -0.62* -0.58*

Volume -0.12* -0.22* -0.15* -0.17* -0.15*

Risk 0.10* 0.04 0.07 0.11* 0.05

*Significant at a 0.01 level.

higherintercept,a largerpositivepricecoefficient,and a largernegative volume coefficient. After day one therewereno significantdifferencesbetweenthe coefficients for the two groups. Adverseinformationrisk affectsthe bid-askspread sincemarketmakersareobligedto tradewithinvestors who mightpossess superiorinformation.If the theory advancedby Rock [20] is correct, then the adverse informationcomponentof bid-askspreadfor the underpricedgroupshouldbe larger,since bothinformed anduninformedtradersinvestin theseissues. Thelarger interceptfor the underpricedgroupduringthe first dayof tradingmightbe explainedby the adverseinformationeffect.

IV. Summary and Conclusions 5Wechose this measureto capturethe daily volatility as opposed to the standarddeviationof returnsfor days two throughfive, since the latter would assume a constant volatility over the time period. 6Volumeis positively relatedto uncertainty,as is bid-ask spread. Uncertaintyis includedas an explanatoryvariablein the spreadregression, so the coefficient of volume will capture another factor related to spread.The holding periodfor a securityby a dealeris inverselyrelated to volume (see Stoll [21]). A shorterholding period reduces inventory risk, so spread is expected to be negatively related to volume.

This paperexaminedthe mispricingof 510 initial A, purecompublicofferingsthatwerenon-regulation mon stock offerings with prices of $1.00 or more. Unique to this study was the examinationof daily returns,dailyvolume, anddailybid-askspreads,from the offering date to five tradingdays following the offering, and 21 days after the offering. Consistent with recenttheoryon the mispricingof IPOs, we ex-

This content downloaded from 210.212.58.34 on Wed, 14 May 2014 06:01:35 AM All use subject to JSTOR Terms and Conditions

1987 FINANCIALMANAGEMENT/SUMMER

38

aminedboth underpricedand overpricedissues. The resultsindicatethat the marketadjuststo any mispricingduringthe first day of publictrading,and thatexcess returnsarenot availableto investorsin the after-market.This was also true for underpricedand overpricedissues as separategroups. Underpricing was correlatedwith ex ante uncertaintyand ex post uncertainty,supportingthe propositionof Beattyand Ritter.The level of tradingwas also correlatedwith underpricing. Bid-ask spreadwas examinedfor both the underpricedandoverpricedgroups.The sameinventoryrisk model was appropriatefor both groupsafterthe first day of trading.Adverseinformationrisk appearedto be significantduringthe first day of trading,which suggeststhatfurtherresearchinto this areais needed. Overall,our resultsenhancethe findingsof earlier workson the speed of marketadjustmentto the mispricing of initial public offerings. We also provide additionalempiricalevidencerelatingunderpricingto uncertainty.

7.

8.

9. 10.

11.

12. 13. 14.

15.

References 1. R. Beatty and J. Ritter, "InvestmentBanking, Reputation, and the Underpricingof Initial Public Offerings,"Journal of Financial Economics (March 1986), pp. 213-232. 2. S. Block and M. Stanley, "The Financial Characteristics and Price Movement Patternsof CompaniesApproaching the Unseasoned Securities Marketin the Late 1970s," Financial Management (Winter 1980), pp. 30-36. 3. J. M. Brown, "Post-OfferingExperience of Companies Going Public," Journal of Business (January1970), pp. 10-18. 4. K. J. Cohen, S. F. Maier, R. A. Schwartz, and D. K. Whitcomb, "The Returns Generation Process, Returns Varianceandthe Effect of Thinnessin SecuritiesMarkets," Journal of Finance (March 1978), pp. 149-167. 5. H. Demsetz, "TheCost of Transacting,"QuarterlyJournal of Economics (February1968), pp. 33-53. 6. L. R. Glosten and P. R. Milgrom, "Bid, Ask, andTransaction Prices in a Specialist Market with Heterogeneously

16.

17. 18.

19. 20. 21.

22.

Informed Traders," Journal of Financial Economics (March 1985), pp. 71-100. T. S. Ho and H. R. Stoll, "On Dealer Markets Under Competition,"Journal of Finance (May 1980), pp. 259268. R. G. Ibbotson, "Price Performanceof Common Stock New Issues," Journal of Financial Economics (September 1975), pp. 235-272. R. G. Ibbotson and J. F. Jaffe, "Hot Issues Markets," Journal of Finance (September 1975), pp. 1027-1042. D. E. Logue, "On the Pricing of Unseasoned New Issues, 1965-1969," Journal of Financial and QuantitativeAnalysis (January1973), pp. 92-103. "Premiaon UnseasonedEquityIssues,"Journal of Economicsand Business (Spring-Summer1973), pp. 133141. J. G. McDonaldand A. K. Fisher, "New-IssueStock Price Behavior,"Journal of Finance (March1972), pp. 97-102. B. McGoldrick,"The Great IPO Battle," InstitutionalInvestor (June 1983). pp. 61-64. B. M. NeubergerandC. T. Hammond,"A Studyof Underwriters' Experiencewith Unseasoned New Issues," Journal of Financial and QuantitativeAnalysis (March 1974), pp. 165-177. B. M. Neuberger and C. A. LaChapelle, "Unseasoned New Issue Price Performanceon ThreeTiers: 1975-1980," Financial Management(Autumn 1983), pp. 23-28. F. K. Reilly, "FurtherEvidence on Short-RunResults for New Issue Investors,"Journal of Financial and Quantitative Analysis (January1973), pp. 83-90. "New Issues Revisited," Financial Management (Winter 1977), pp. 28-42. F. K. Reilly and K. Hatfield, "InvestorExperience with New Stock Issues," Financial Analysts Journal (September-October1969), pp. 73-80. J. R. Ritter, "The 'Hot' Issue Marketof 1980," TheJournal of Business (April 1984), pp. 215-240. K. Rock, "WhyNew Issues Are Underpriced,"Journal of Financial Economics (March 1986), pp. 187-212. H. R. Stoll, "The Pricing of Security Dealer Services: An EmpiricalStudy of NASDAQ Stocks,"Journal of Finance (September 1978), pp. 1153-1172. H. R. Stoll and A. J. Curley, "SmallBusiness andthe New Issues Market for Equities," Journal of Financial and QuantitativeAnalysis (September 1970), pp. 309-322.

This content downloaded from 210.212.58.34 on Wed, 14 May 2014 06:01:35 AM All use subject to JSTOR Terms and Conditions