Pe OE A ON EMS Ce neti g Hn He! ww (eth et VTE POCWIN Ded ty ELI out ete Ff MOY 55 nT Give ttt Malta aves 02 GENS EIN Mf BY PRAMS M GA Da Oe. ee AEE ON hed y “yy ais 10) 20 yi te ah 9 y Ao PEN ey ea wa da SAMA aa beh Lid POt ye tg ha tere Hd CRNOAN Heya gn Marans Teg MDa tebe yt Leet’ Sena INO Rit AMAA PME ENOL NA ites pHa i , HE bya oe ulby pe i fests LAS ‘ay Wa en Sy AMEE GOED Deweae Bt wl ga MAI se) WME ANN OL MOY Loni often gery ate *yieeet ue SEM bye NVR MeN RY Ulby Fee abt LiinalarsaBa yn pee Whee Aer WI Pee (mew { - * 7 PAYEE a Papeete 1 20 0 cen oe Wat Pg 4: Nau , Maal Ma ee > 4 yiel siCody BAND Ry OLIN EB A (tie tab Yah ee Nugiet Te EYP AEN pay LIN ny aay NY bw aifae eile nba ha Wa Be Atytte eh DE er eM i nL Sa oh ny Ve Preis eet AN Bh oahy ee A ediatig qin iW wha tte itt Lehewae wo Hei " Hannes Vea S WERE penta Ca Meccetletng thy we 6 COSY ER pha, Cara rane Mw Cron maiten Wart sre CON UL ENEWS ON sed nay ri PHENO aT eS OTe te weee nh tn pe by Dene SNA oy iy ae AWAD ea aye VAD Yi ony ts Wiehe ete ALAM EY, y AN Peat ing ‘yt 4 ’ ea ts NF Vt tae ae Iris hadi fre tye yin att ane RINE SAL RW CCEA TENT FOI ND? Aalto esa bang A reulhe vinta, Oy pnb LN Aadnod ae thy Hing WU Pat pat Hla Og Sa ty SA ME BAR toe Vet CLIT NOMS te fee at OME I aes he Mele erated sibs neat uf Pela rN Anette ig ask adept estan FD Pe QUOTE OND ha ein Ce PLAN CULL TS Itty Het COUN. ALU eH OFC oe siNgte LIUPSINOY KUMI EO OPV Op ete dla Aen tee EBM Neues Shred ba DYRMD ATER TALIS fi BHA eT UE Lp ey Mette a ie DS Ae Meee ated REE NN Yad aw Ho donde Om Mad te fle MEV VE NE LMT 8H, oon ZNE de ENGNG aa x SEMEL IR Ly VEN RD ND AMDT NT ESE GRY Hy AAD REL DS Ht ORE RSTO et ee ee er Liar Wy eR ey Med Kar ver ee oh Mth Me tee! pL Li Mesa ten PNAS tie tial ln dy Fe SPEED ODE UME et edt ot! pith Like Oopbanys y pes & Ue UD Aten te pivot as anal eC RT a Sa Pe OPEV hy OS thw AeA EN GA TUES VEE IN pont aN EI 9g 6 AN BO MTA PT EE EME fie Me EN | aie" CO aU he ae Ce en Ce Mattes Cadas ADO aE We aE ayes BEANE RIM UY HUES eh bey ty r TOV NWO een hee thai Rad a sha ath ye xb sahil WA ed erst CP yiea da ot f fant pone EER Sy ind + aol ar NV iawn deel Ae obosarrahet lay gb, Pn ba Drie tert yi PU Wes Hye wei] am Met py Leo On OU ie a a Sh OM aye whin wer ryt, PAN UN Vite woh thas Stan ey illo lait Lee aa a aS SU Vip ARE AE Hie AN OY Sade] AACA EGE Wey eee eae Tris fear RHE LYIS WOE LAYER ¥ PYF WIEN SEN De ty Md OVany ge Letty SALMA Te gare E AUT ENG UH EYE LU HL tye y OP thw ba CRIN WS Pe AMAL A ARONA Hg Oe SAO FOE tor A Sg aS byisae WOM Leche is Cah ay in PE pa rmenet ee agony ty M85 5 > yh ys ay; Soy Ys DY balan Neue yh Taare W ce is WAL VE NIH ANT LE OURAN te VAIN NOD Lt VAI AS OLE YT ET SDM hays BENT fa yer edta Aimy aye Yuedaa enim eis f a ‘envied PA cK HU pee CLL CALTL ean Ne Rete Tien Rd Oe Ts OLY OME Lye ‘ Site IN ae ie ox Wen ee Onno BAGO MATa yph ule TN RE Hey tu SN TET Sabie ih an AON TEMES 6 NENA VAL SPE MEAG LOIN AM pty ty arrers oA ysy hs Definite ZW OVER IS YAS Me aIED ET? eat ReLevy pyitdbaon ce view it PUrkriurue ee arty, Poel Wig te Pah Pde BOL Ute ABTS cel gates SeieaM yl 8 Di A 2th Ad Aydt Nana APA A Noite eda ms MeGM EELS vena y ieee Nan ae PARES L NED OL Ns SANS Ltr ‘ iby Sa oars es bs wre ; Pha INS, ANE BT MEL 9 DUAN tie Ly Vd Ve PNY VET SFP Liye Ly sho pe 1a fely yee Lena rx cherie win pat ¥ fess MYR CSY AUS UNG) ety Fa SQM tare wih govt She ge Ue fe A ALA Lele fe Wien NO i Ed (UR tibs ae n cette STD Eee MODY we VN OM Mees oy Pat Sa hI WD og. tS BY ab La lar te Maeno LON VND NT Bytintid iy apa WAN Ve AF ee FOL Wace MEA a ES WaT EM Vaiviretnn ng ninth fligdtin t's POI OU EL Mt ed Me SLM, BEAM LAT PAWEVRTA INI EH EL fe Lead ENS! vy MANS DRE NOULY hae Ube ygGvane Can § awa? > ee TRY THEN A yy AV en a viene Dee UU gn bth HE GL Ne ga) Baa Ryedale Cage eats Mendes Mpc agiT chew Ue late lI p ry nomen tak OL NEN tad Tent eileen iet NAY 0 Aidondin Cuswrd yop itent Feely rowney hy; manly’ fare wien ei Beat ens ge EYPLC 4 mV SOCOM LUN BAT Me Cie open EET Oo NiowM 110 Siu [Psealnd sind, a ety PN orp Mi dteh cr” ald PROT GN NS Rea AAR Ce OEE I Pr Bias er a BORAT DD FoR teed! rit A et eh wee Fed WET OG a wy MEUOMION BAe Sine ote Tele FY Pe matty . ANE Steg Selene lCH ar anOrN um vets Le ay Ma fa idle AM VEN AERA ee wipes eee etl VM AD EME NN SYM Sy by ig wath! tala stare LV ENEMMR™ CDV LES LIN SN fe Bn Ne Ae chee ea eet Stk IN MOL SUDA Ts ge ath st we hey HY ww till Hw debby eo at SEMIS VSIA, te Fos oe Readiness tts Sra (Vals ®APMINE, . Sh iLO SY Ming out yh, % ‘ A STUER STN Nails e ALPS VEEN UNS HERS eM RR pee N Oe APE teu Ca Pee aay hese pnt ED Salve tw kam rugs Pedy Oy Veta eile et te WEA MY, thd A sey Py CULMS Dip hohery 15 Mad Pa art ae i peeves, tye CU eae ea ny the meme tid 4 FUMIE Ne ee AN Ra, LENA adage ey ha OBEY OV 7 RAY eI8N he ncm 5 OVE Le pede x hil YE INST POMEL TD se Ma Sa PS eta yey oa Nad arse Vt Vi euiialy a EEN STS yey are tks % Be Mien, Mac tved ent sia Re Hythe Miho La 4S Wy vanes bere ats ry uae Nal bar alent ahi MD arg HUN SY ERK NONI gw 9 we Rene ceaecattudl Aeris heh tahoe Fel Epcot ate “deen imadsibinabaser denanie eine Ane Hat itavtatig etal tera itia Patan aaron (intial me wey Te Mat COOW Met L AM WE RTA singin een Feu INU oe en ere LS SSN ovate encour Te peterennnnennerien SANE tA Rebate CA La ae ae Lae i CAEN ERM Radin ine alight, Sete MA ae Waid A we eee LY er TEE WTA eA at's WEE DYE Ly righ 7 a Ee RN Lise OCIA TALE DIMES IN ATs fe de AAd 6 Pattie Tye gs END Ted Nt Sve hana oat tA ie ha irae NGA cae MARE a Gy ey Peele snl eeayn ahs WR apa tintin ‘ Core I DEEN WE Ok PDN MO ORR PEPNOO PWS Aho Sh rete ve MARINE amare Vac AR Re ee eit VAs Aide ery wp ose PeoA MIA De oe hile NeSt y Pe ad Cay Aa HES Siri Nel Ade TNE OL aN ARTO IU gee SVE got natel OI ERIE AEE OL 08 Le NSA RIVE TEN Sot hy Ld ie VIBE ROD GAS TAT AYN OY f wi LE fp pie Clee al etl pltiedi arte OM ey [oes BAAS HEY WM MUNCH PLT ici Med cet ey tia ae A “ “ Cay ge ern a ae eer ree Leas etal wa Be To) or sia tps Mentos oe Ye Piet 1nd MN WT Rs SGD BM a Gell 9 Fy Fay WO DAD Ser ee aye hehe Nowy 8% MRO Dy ¢ MMA ONT OMS OLE VAD Cnt eae PEEVE ie ot Vt Fee Ch CA OP ORBEA VED Lp SIE tha yee IN We De Muha talt We neoied sly 2 tether atone yr e lad: Dh en Cal aK Molt Ha CllV WOE Ey een ge 4 Fad dot Adie hth te PWN TNE BP EIN Vary ey Neg Pye ewes Ph AL LUT SA INE TEY Fis ULI NS Gllad G AQE OMY CAL NOE Ae Foe wee indy FT? VEEL eR abby Phytepey on Sree gs LOYD Be OR Mae gh 5 tay LEP bo ‘ d DYER AND Hes oyytnpy oN Sra List hey SA SO sie a Neciaeey its Vessco ; ve pace Geet " bh t Re NTIS AES TIPLE Wee NE RENG MLE Y TW Mbt sD, . 4 MMe el Uta EI a) pak etal bie Ta Av Pe! SE AIM ATO oan Mh Mee ate teense eh OM PRR ee WANs uita Py AIREY Ny PY OND MINING DS fad Ne atid, “Dgity # ne ISIN Tad 14N ARGH Pugin s Ope Mensinn 2a 4 swt m Cara AW by er yy WAIN Ade mr is fed a rigeretyed eH yy fer NA eee at AM wi TEN TO He OEE AUTTRIN Ha PERM ERALWIOZNG wd top yan fee PAVE OW ot WOM aay SII Oe emw an ease Bie tA gyi LEV ITAITH PAI ont UF DY Fe Phd RIE way sue ameny a BEERS AY tev es crater ae str e O OT yheow eae Bgl pital Oe sata Tee Mey g iNet tl att Maca tee Be HA ea al WTEC UN WW Waa nik . WEA te Sern art ied Une thie pe SEL P NE WOR ea Le ha eee Be Me TNA et ante EU OURO Kivi yeety Neve yey 61S wig? nA GAY Wit NOE, IE MER GEE HEE as Wire PAA NET 9? gdIV OM eater PRE eae ty Ved IMEI TENE EN ye ny Ser ern CDE acer cart ELIE ee thu Yat ey UE vereneenen Vw Lah Meh te hy uy yey ve Nid saw Net aus ea i ee UO urtanr oe it ar re Pe te Air rt net rn ips Ne 4 UB TOD MM NILES et Waele Me Ge) Magy, Wee She ee Rey ns yee Age be ye GWirhoneeee RT er Tat iN Rae FE at cny MY whey we ey tay Nee Co tet Cea Le Aah? aE SAFE HV ENE N wod sapn yh AN Abele SNe ANT reat? at afl ested ed Pen nn Cae DET AY ag Nig ae pi Poy be fe RAS VIR PVA HEAR LEI pty git Hep tes reste ais Wares TAQ omrwthe rem! arte Ais DOME te as eps : Shige J? PATA gf VM ULTRAM Me ee GPR ANL ay NeW ee rE RC MEY y be byibyihy'e Nalsine™ oY pAmekaein lA eel: alas 2 ‘ 5 ( Tee ng Ve PA Ea ged My YW™ WN POW UD os 2g at COT Ih ypu 6 : < OMEN wid 9 Tons SAIL AN wi MB a Fan - S FEN in VIERA TEN BEN LEVIN Le Oe 1% CEE ae ee es a alas te SR eA Ta eae ee AC anenewboanyn oa FS SO NSU RN ees WEE ON nrecene ee iEe TOS LE Loe i8 ine ee iA OL OLIN E! Wethe eee CER EASE VAR ew ty arom SIDER Is cartons ene ‘ vos 7 : Pe RETO AE pe et kA et CE eg ged. PM a8 ec TOT Mad ed nA le EE BEAL EINE EN Sa? el ork ni ae) Se Ce itt naling dia aie UNS ea Pa SPIE my te eh Nd Piel SNS Deter Raye ew MG Myre ak ye ot VOM D SPER vere rem stews tg Oe P <4 bye rey ery “ PRPS bh atlas My be ey ve INS ¢ ES Me Iie Hse NT va. Eh hed el Me 2 Ff aS Cate Tene wren) eSlae pop eea ane oy VIN Nas ad aa ed ay at obit tens on ed UTE Sy OP Pay sir seen dye Seat HOT NEES ETA ERIN O MOSES RTS Large ices , ASI rhe eg : : S/W Nel bITa A NYA rm MENS eee aon tesat 11 fig ieee! att Jes ed 4 ae NAO tie ts > “nef PUAAEO pr ere) ee ieee! SIME yt yeti 4 EU oat Va Ee wk bia | nA eli aatite aes Pee ‘ iu, “ cs Seay ete ed PINEDA YN od by Nowe : ee ted ST pty beat A aU te ete hue Devt Fiat Ms att NOD HET AN Lod a thy Ne foal? st Servet te EYEE Ft dae AE AH MW lh MET IM YN Sogn My, fete eae Feel, QTE Ot bd aah ene y POP ey > Fab tale aed Maile BOP OME OTe roe bet ODF Em LD Mot Roe SLANE ys ee pane nel y! Sd WAIT WA. Onetn ms onde paints Ruta esa Casa aah ewe : ast a pee NF ate Ae 4i2"y it tba Sy 0 nA. 1% 6 eet ye staat x . Bed Saas Ph resales a> Hite Mee pve SINR NS SITES WO Ny WN FSI RA fest VEMCENS Se ned Wes Me WANN TD Tf ea MSS Nie Soe eri Oke Oa ri SWE aN SUE wpe telboys meer WIS My henry Leh URE SIND 9 Saw whe Hints inves etrbwe ree AN Ata Loy 9 Yacht aM riety ONES Sa rem TSUN Hey Pe ede hare FOL ONE gH NESUL AMINED YN Py MTN Ey ayiy ww Se fed Oe sa etn ely, Neate whe FIG VE MODES Cnty TWN H ytlag oy Cy ener eee SPipt hn toa ER rep UNE ye bes * ge aon trae Anab wi ovis cre ALA 95 oe . : vriins ek hg NY MEIN Poot thas are © aya vt Leas a Rah AL Pere atl aad MOS waded OMAP LUMEN ROIS ES ES WES Oty BE Ty LeN SNH Ty NaRE TY Wis NyiND athe MEME hs RS Ket ote Ped aod XS ay ‘iar agar dace oe Ga al VN M earns Dv encne aris ran . PtP e008 AMA SD " 0 oe PONE AS WN tatty JG ate os ~ Ye ee tg a nae eres mee a Sram es TopMia Va Wey dan ake Pi a Pe ae An eTae ot hese cata bean ot " Ce al dete Le ree Se NE on awe bay sve Ae ee La SIDE RUAN SEN tg ce SL TEIN Ce weg Oy Nive soc by “aun taal FP eqreynys , WES FM HN INI SAIN be. EM ali Hoe, faye Sigh. Cintas “wy Phat wey wlan YER gh 9 Sie en Hg VET SN ley ety WAlen erin nee y png Fd a Neary hy, RANE BIN iNet sy eres aes Sei ht ee WE ag tT Speen FEM OT arma ey = : Seca! Pe are et re POTN ere whlbe Terie es Bem yee mn Pe aa I oe a ty an ened Min ABN os eed Tae mg tym TD Tad gh Sg Rm areas ter We sae VA A Pos wynytany lee BAM ey nk? Pee ene oot SM Na ie ¥ wy nm ae Aine ed oy Wee Wy) ye A a a ‘A nes ie) st et Fa Lat lec OLE nN: a. bw a) H) Pde p Wis i as A yl He " Ae PSS > ' ; 7 . a fi nT Ai SM a ia see sh Gere \ y ay. ~ q cs a \ a & . sé Z be a ‘ E J Hy ; Fi - 7 e oe ( ge (i? | vit | VOLUME 82 ! Number 1 J Our nal of the March, 1992 WASHINGTON ACADEMY ..SCIENCES ISSN 0043-0439 Issued Quarterly at Washington, D.C. CONTENTS Articles: JOHN H. PROCTOR, “A Theoretical Basis for Intentional Organizational Change with Comments from a Thirty Year Perspective” .................... ROBERT GOODLAND, ANASTACIO JURAS & RAJENDRA PACHAURI, “Can Hydro-Reservoirs in Tropical Moist Forest Become BMVALOMMEMtAlys SUS CANA DIS? mis tions eke eek oeccis oe cis bickoks & eiclevone igre eioeioece 19 ROBERT GOODLAND & HERMAN DALY, “Ten Reasons Why Northern Income Growth is Not the Solution to Southern Poverty” ................... Sy PAN SELUCHONSItO) GC ONTMOULOLS Ge esis ete Fe ree aires RR hae eae Washington Academy of Sciences Founded in 1898 EXECUTIVE COMMITTEE President Walter E. Boek President-Elect Stanley G. Leftwich Secretary Edith L. R. Corliss Treasurer Norman Doctor Past President Armand B. Weiss Vice President, Membership Affairs Cyrus R. Creveling Vice President, Administrative Affairs Grover C. Sherlin Vice President, Junior Academy Affairs Marylin F. Krupsaw Vice President, Affiliate Affairs Thomas W. Doeppner Board of Managers James W. Harr Betty Jane Long John H. Proctor Thomas N. Pyke T. Dale Stewart William B. Taylor REPRESENTATIVES FROM AFFILIATED SOCIETIES Delegates are listed on inside rear cover of each Journal. ACADEMY OFFICE 1101 N. Highland Street Arlington, VA 22201 Phone: (703) 527-4800 EDITORIAL BOARD Editor: John J. O?Hare, CAE-Link Corpora- tion Associate Editors: Bruce F. Hill, Mount Vernon College Milton P. Eisner, Mount Vernon Col- lege Albert G. Gluckman, University of Maryland Marc Rothenberg, Smithsonian Insti- tution Marc M. Sebrechts, Catholic Univer- sity of America Edward J. Wegman, George Mason University The Journal This journal, the official organ of the Washing- ton Academy of Sciences, publishes original scientific research, critical reviews, historical articles, proceedings of scholarly meetings of its affliated societies, reports of the Academy, and other items of interest to Academy members. The Journal appears four times a year (March, June, September, and De-. cember). The December issue contains a di- rectory of the current membership of the Academy. Subscription Rates Members, fellows, and life members in good standing receive the Journal without charge. Subscriptions are available on a calendar year basis, payable in advance. Payment must be made in U.S. currency at the following rates: UiS..and’Canada «.: : ):a)ee eee $25.00 Other countries: 2... 3: eee 30.00 Single copies, when available ....... 10.00 Claims for Missing Issues Claims will not be allowed if received more than 60 days after the day of mailing plus time normally required for postal delivery and claim. No claims will be allowed because of failure to notify the Academy of a change of address. Notification of Change of Address Address changes should be sent promptly to the Academy office. Such notifications should show both old and new addresses and zip-code numbers, where applicable. Published quarterly in March, June, September, and December of each year by the Washington Academy of Sciences, 1101 N. Highland Street, Arlington, VA 22201. Second-class postage paid at Arlington, VA, and additional mailing offices. Journal of the Washington Academy of Sciences, Volume 82, Number 1, Pages 1-18, March 1992 A Theoretical Basis for Intentional Organizational Change with Comments From a Thirty Year Perspective! John H. Proctor B-K Dynamics, Inc., Rockville, Maryland ABSTRACT Application of scientific methods to organizational behavior followed work done in the military and industry in World War II on team performance, group behavior, and organiza- tional design. A new scientific discipline, organizational psychology, grew to fill the gap between psychologists studying individual behavior and sociologists and cultural anthropolo- gists interested in large group behavior. New approaches had to be devised for applying traditional applied research designs and measurement techniques to the formation, growth, success, and failure of human-centered organizations. At about the same time, a holistic or systems science was emerging. This science, when coupled with the tools of conventional applied research, brought a fresh new perspective. During our lifetimes we have come to view organizations as living organisms (“living systems,”’ as James Miller calls them) that grow, adapt, evolve, interact with other organizations, self-organize, self-regulate, and, based upon feedback from the environment, modify behavior to succeed or fail. General systems and cybernetic theory provided a framework of constructs, models, and points-of-view that was helpful in organizing and formalizing thinking about intentional organizational change. Using a general systems paradigm of Organization (O) Relating (R) to Environment (E) through Time (T), this paper both discusses ways of first distinguishing and then synthesizing structure and process and articulates criteria of organization effective- ness and efficiency. It then suggests ways of synthesizing formal logical problem-solving and human problem-solving processes in the context of concrete situations faced by organiza- tions. Such a paradigm serves to help clarify hypothesis formulation and evaluation using conventional scientific methods. However, the methodological focus on organization objec- tives, goals, and missions has revealed to me that individual and collective options arising from values lie at the root of organizational effectiveness and efficiency. General systems/cybernetic theory can provide useful scientific paradigms when coupled with those of specific scientific disciplines, common sense, or intuitions, in the sense that it offers an interdisciplinary theoretical framework for analytical investigation of highly com- plex human-centered organizations. However, there are bounds. In any real sense, scientific investigation of human centered organizations cannot be value-free. The general systems paradigm cannot, of itself, explicate root values which arise from somewhere outside that framework. ' Based upon a paper, “Management Systems,” presented to the Mid-Atlantic Conference of the Society for General Systems Research, September 1973, and an article by E. J. Burns and J..H. Proctor, “‘Sistemi e Administrazione,” appearing in Paradigmi a Societa, (A Systems paradigm), Franco Angeli, Editor, George Braziller, Publisher, Milano, Italy, 1978, pp. 87-101. E. John Burns, my friend and colleague, and I discuss these issues to this day. 2 PROCTOR I. Introduction Contributors to General Systems theory/cybernetics have increasingly recog- nized that despite the advances and the useful insights achieved by theory, the nature and role of values remain paramount in every intellectual construct. As Sutherland (1974) states: . . any information any scientist acquires, by whatever means, will owe at least some portion of its substance to non-empirical, non-inductive imagina- tion, such that a value-free science is both logically and practically impossible (and perhaps not even desirable). To reach a similar conclusion, this paper describes in some detail an application of general systems theory/cybernetics to organization improvement and dis- cusses why in human social systems one always faces values that transcend and direct the construction of all formal structure-process sets one usually denotes as “system.” In my experience as an organizational psychologist consulting with clients in government and industry, a large number of problems and situations are seen more clearly when viewed from a general systems/cybernetics perspective. I am a technologist who attempts to apply that perspective whenever possible. My work, performed in a highly pragmatic context rather than a research or aca- demic setting, has led me to certain conclusions concerning the nature of general systems theory/cybernetics as a paradigm for scientific approaches to under- standing and organizing human social systems. Chief among these findings is that human social systems at any point in their multi-year life cycle present a hierarchy of phenomena that, while inherently interactive, has a precedence order of relationship and influences among levels. John Burns and I have been asserting for over twenty years that beyond the familiar systems concept of levels within levels, there are qualitatively different levels—perhaps we should call them Levels—such that there are not only levels within levels, but Levels above or, better yet, Levels beyond Levels. This result is suggested when one poses the now familiar question: ‘““‘Where do goals or objec- tives for management by objectives come from?” In attempting to answer this question, we view human-centered organizations a little differently, through a conceptual lens of general systems/cybernetics theory. The result of using such a lens is to try to step beyond a description of the separate notions of structure (management systems) and process (systems man- agement) or the semantic distinction between systems of management and the management of systems. However interesting and compelling the extensions of ORGANIZATIONAL CHANGE 3 these notions might be, this paper explores the interactive connection between structure and process in terms of the missions, goals, and objectives of organiza- tions and the measures of their individual unit and total organization effective- ness and efficiency. A General Systems Paradigm In the broadest sense, institutions, companies, corporations, agencies, and churches are viewed as “‘organized complexities.”’ (Weaver, 1948) The orga- nized complexities are certainly systems according to Bertalanfy’s (1956) defini- tion, “‘sets of elements standing 1n interaction.” They are complex in the sense of Simon (1956): “Roughly, by a complex system I mean one made up of a large number of parts that interact in a non-simple way. In such systems the whole is more than the sum of the parts. . .” In terms of the elements themselves, two basic classes can be distinguished. First, there are those elements united for a specific purpose in a controlled and bounded arrangement, as distinguished from those elements that operate out- side the purposes and the boundaries of control. These internal elements (with bounds) can be called collectively the Organism (O), and those supportive and constraining elements external to O can be called the Environment (E). I ac- knowledge that we are on dangerous ground when we select and classify ele- ments, for as Hall and Fagen (1956) pointed out some time ago: “‘[I]t is no mean task to pick out the essential from the nonessential; that is, specification and subsequent dichotomization into system and universe is. . . a problem of fundamental complexity.” Stafford Beer’s “‘viable system” contributions di- rectly attack this problem. His thoughts on “Divisio”’ (1960) are particularly relevant. The interactions, of a dialectic nature, both among the internal ele- ments of O and between the internal and external elements, we call a Relation- ship (R). At any given moment these relationships have a present and future weighted existence which constitutes the Organism’s actual and preferred states in Time (T). (Fraser, 1975) This thinking about interactive relationships among the internal elements of O and between those elements and external elements of E coincides with an analysis by Mark Braham (1973) who, in developing a general theory of organi- zation, has perceived that the fundamental process “. . . is cyclic, involving alternative periods of divergence and convergence.” Certainly organizations develop structural subelements with differing func- tions, and considerable individual and collective effort is expended attempting to orchestrate these different groups so that they will play together within time phases and over time intervals. But interaction between divergence and conver- gence also manifests itself in particular ways. It arises because of the practical 4 PROCTOR necessity to make the heads of organizational elements responsible for produc- ing at maximum rather than optimum levels. This is a practical necessity wher- ever optimum levels are not explicitly known, which in my experience is gener- ally the case. The interaction also appears as the conflict between change and the status quo. A particular manifestation of this dialectic (for that is what we are really dealing with) is encountered in corporate planning, especially in advanced planning, where a means is sought for performing the transition between present and future states of a company where the guardians of the organization’s present state can be distinguished from the exponents of change despite the daily singing of the corporate anthem (Morris, 1974). Improving the performance of a system or parts of that system begins with either the acceptance of the client’s organization (O) as given or the articulation of O in the client’s terms, specifically welding the one and the many into a dynamic whole which has both individual unit flexibility and holistic cohesion. Then the environment (E) is specified. This is usually a problem at best, but what is especially intriguing is that case in which the basic character of the company itself is the subject of close scrutiny. In these cases the basic question surfaces: What is the function or mission of the total organism? This question must be settled before the relevant bounds of E and the relationship (R) between O and E can be determined. Answering this question is by no means as straight- forward as it might appear, even for consultants and clients working within highly structured societies that encourage such well-defined institutions as cor- porations. It becomes an extremely difficult task when dealing with governmen- tal, educational, and other similar institutions which operate largely within self-interpreted and changing boundaries, established and justified with too little debate and fewer empirical facts and measurements. In essence, I am employing an abstract and admittedly idealized general sys- tems model—that of Organism (O) relating (R) to Environment (E) in time (T)—to help articulate and implement with a client an interactive divergence/ convergence process which will achieve optimum rather than simply maximum states of output. The next section describes the O-R-E-T model in more detail. After using this model to approach organizational development for some 20 years, | am persuaded that despite appearances of sophistication, we are shortly led to certain hypothesized assumptions which cannot be deduced from the O-R-E-T model or its method of employment. These considerations form the basis for the remainder of the paper. They lead to the conclusions on general systems theory/cybernetics not only as a paradigm for conventional science but also as a currently conceived vehicle for assisting organizations in solving prob- lems of efficiency and effectiveness. ORGANIZATIONAL CHANGE 5 The Model In the development of the O-R-E-T model, major emphasis is placed on the relationships among and between the constituent elements. These relationships manifest themselves through the statements of goals, objectives, and missions of the organization. It is the articulation of these relationships that is at the root of this approach to management systems and systems management of person-ma- chine combinations in human-centered organizations. The principal reason for using this approach with clients, stated in its simplest form, is that relationships among elements is the essence of system. Unless we focus upon and build upon an understanding of relationships, we can have no concept of the cohesion underlying the convergence and divergence processes of the many organization elements which are required to operate in a unified fashion. We would fall prey -to maximizing the product of subsets, rather than, as far as possible, optimizing the outcomes of the whole. Working with a variety of institutional management forms, I have observed that there are a number of subsets of attributes within O, 1.e., organisms within the Organism called os, which stand in a variety of relationships (rs) to different aspects of the Environment. These aspects describe E and are subsets of E, which is itself a set of es. These os within O have a variety of relationships to one another and constitute elements of environments of a second type, internal to O but external to each o. Stated simply, these are intra-organizational relation- ships. Hierarchical structures and processes, over time, could produce innumer- able distinctions between internal and external environments peculiar to any of the os. In other words, a particular subsystem has relationships with other sub- systems of the organism as well as with those attributes of the environment that lie wholly outside of the organism itself. Recognizing these orders of complexity and choosing to work with a whole institution as the O, one is forced to define very carefully the boundaries or constraints upon the activities of O, which involves taking an “‘outside-in”’ view of the particular corporation or agency.The first order of business is usually to assist the client managers to identify or define the system of rs as a set of pur- poses and related states to be achieved by O with reference to E. First, we define the overall R in terms of what is called the mission of O, which is O’s own guideline for the type of R over time that client managers would like to have. This mission definition results from the best thinking the client can do about interactions between E and O, making assumptions about both. Since E induces constraints and even a set of values that restrict the “rationality” of O’s mission, the consultant’s job is to test the assumptions about E and O logically and, if 6 PROCTOR possible, empirically. Client managers are presented with decisions or trade-offs, in which choices can be made about the kind of Rs and Es desired. This is what happens when we arrive at a statement of product line, market, and raison d’étre (such as maximizing return on investment). In effect, these statements have put in a general way the combination of desired E and R, and assumed much about what kinds of Rs are possible, proper, or good and what kinds of Rs are impossi- ble, hurtful, or bad. A pragmatic mission statement is developed in terms that facilitate its attain- ment. This task requires the client to construct a set of sub-missions, goals and interim objectives that permits a definition of the organization structural ele- ments through which objectives will be attained. Both structural questions (who, where) and procedural questions (what, when, how) are answered narratively and graphically in terms of necessary primary functions. The functional net- work of subgoals in essence defines either the structure (though not necessarily the form) of the organism or its constraints. It does not necessarily bound the possible process alternatives. Since we are attempting to specify a goal-seeking dynamic condition over time, we speak of goals as configuring future desired system states and objectives as configuring immediately desired system states. We are essentially constructing a model of interrelated goals and objectives, a 2(r) system which is stated in such a way that it defines, at all levels of aggrega- tion, the desired r with a specific e which concurrently ensures that the rs are consistent with the overall relationship of O to E. Such a structural description 1s called the functional structure. In recent years, various organization develop- ment (OD) techniques have been developed for setting organization objectives. The methodology is generally referred to as management by objectives or MBO (Beck and Helman, 1972; Roeber, 1973). Functional Analysis Within this functional structure 2(r), we very carefully distinguish functions from activities. Activities are the mechanics of function performance. They derive their right to exist only with respect to the efficiency by which they allow the function to be achieved effectively. Functions are goal- or objective-defined collections of activities and the functional structure articulates the web of in- terrelated goals and objectives. When involved in developing procedural, organi- zational, and information solutions to management problems, we first devise the functional structure through functional analysis. All activities that are not amenable to goal-oriented articulation in a network culminating in and consis- ORGANIZATIONAL CHANGE a tent with R itself are carefully excluded so that we can devise measures of effectiveness for O and each subsystem o. ! Such a view provides a perspective for operationally distinguishing between effectiveness and efficiency. Effectiveness is defined in terms of the quality and correctness of an organization’s stated mission, goals, and objectives; it is a measure of their relevance. Efficiency is a measure of the cost per unit time of achieving an output that constitutes achievement of the mission, goal, etc. Hence, efficiency can be measured by itself without respect to effectiveness, even though such a measure is surely restrictive. Effectiveness cannot, from a dollars-and-cents viewpoint, be defined without reference to efficiency. Given limited resources, an organization cannot be effective unless efficient, but an organization can be highly efficient without being at all effective. ‘However, the operational problem is to define both types of measures and establish their relationship. By defining an organism O functionally in terms of r, we have constructed the means of defining measures of effectiveness. When measures of efficiency are developed, the two types of measures may be joined into an operative scheme for management. In my experience, the failure to approach the problem from a functional (7) point of view is a major cause of untold confusion in institutions of every kind because the distinctions between effectiveness and efficiency are not clear. Further, the use of functional analysis as a means of capturing divergent and convergent interactions 1s fundamental to a correct employment of general systems/cybernetics models for the management of organizations. To better develop the idea, let us use an example of a function. In a large transportation company, maintenance is a complex, important, and expensive organization subsystem requiring a large staff and an equally complex organization. The transportation company has labelled major elements or subsystems as produc- tion, production control, engineering, reliability, administration, planning, etc. Each major element performs many activities. However, one of the company’s ~ overall objectives can be defined as the concurrent reduction of maintenance costs and improvement of safety and services by means of the modification and improvement of spare parts design. As such, this desired way of relating O to E is a function that induces a convergence/divergence coupling of all the above organizational elements of O. That is, the role of each major element in this function clearly differentiates certain elements of the organization from others and at the same time relates them in that function to the overall organism O. In other words, the whole point of the 2(r) system defined functionally is to con- struct a holistic description of all the functions of an organization such that each is related to the final mission of O in a clear way but they are carefully distin- guished from one another. This same form of description unifies the various 8 PROCTOR activities under the functional objective, facilitating review of the necessity for this or that activity to exist, and thereby distinguishing between issues related to the efficiency of an activity and issues related to its relevance to the whole, i.e., its effectiveness. As shown in Figure |, from a methodological perspective, we (the consultant and client) are beginning with some configuration of people, equipment, facili- ties, material, and money. First, as mentioned earlier, we attempt to define the mission and to specify or bound E and R. Logically we would specify the present E (E,) and the future E (E,) as well as the present R (R,) and the future R (Rj). Often these are not compatible on the first definition cycle in the functional analysis and some iteration is required. Next we decompose the E (both E, and E,) into various constituent elements for which some set of relationships (rs) are required. These rs define the objec- tives and goals of O. We group these rs in a manner that we hope will facilitate a definition of the processes by which they may be performed. Such groupings vary from case to case. Their formation is largely an art form, although the O-R-E-T model provides us with certain guidelines and checks. These groupings are examined for interrelationship: How does the achieve- ment of one function impact the achievement of others? These interrelation- ships determine the network of rs (in terms of goals and objectives). Finally, we construct the levels of feedback and aggregation for the functions. Levels of feedback are defined in terms of how the function is evaluated and modified. Levels of aggregation specify subfunction or subgoals as objectives whose ac- complishment determines the achievement of a larger, more inclusive, goal or objective. Functional Process The next question is how the functional structure is activated—how it is articulated in terms of process. The dynamic aspects of a function are viewed as process, a transformation from an existing r to a desired 7. From this viewpoint, structure may be seen through time as stable process. In other words, if processes exhibit some critical degree of stability, they will form an observable pattern which I label structure. The key to identifying and examining process is the specification of the information and communication required for an O to pro- duce an output. It is by means of information that processes are initiated, sus- tained, and redirected and that objectives are accomplished. It is fundamental to describe the character of processes from a mission-, goal-, or objective-oriented point of view. As Sir Peter Medawar (1973) has remarked, “‘[I]t was not the ORGANIZATIONAL CHANGE INITIAL CONFIGURATION OF PEOPLE, FACILITIES, MATERIAL, MONEY tnt tee BUR FOSSA SELECT ENVIRONMENT PRESENT (Ep), (Ef) CHECK FOR CONSISTENCY OF MISSION & ENVIRONMENT SEE OE OOO ig OP LO NETL ELISE REVIEW e's r's VERSUS E Qe Qo YES FORM FUNCTIONAL GROUPINGS, RELATIONS, LEVELS FUNCTIONAL STUCTURE OF THE ORGANIZATION Fig. 1. Development of the functional structure of the organization through functional analysis. 10 PROCTOR devising of the wheel that was distinctly human, we may suggest, but the com- munication to others, particularly in the succeeding generation, of the ways to make one.” We initially present information descriptions in terms of transforms. A trans- form describes the logic of a content or meaning change in information. Then we develop a description of the information transfer. A transfer involves the movement of information-bearing symbols from one place to another. A de- scription of the flow of information in terms of transfers and transforms, within and across functions, within and between levels of O (indeed throughout O), and between O and E through time (T) completes the functional process description. In practice, available time and money prohibit such a complete description. We can now examine the existing elements of O and our initial configuration of resources (see Figure |) to determine modified or new configurations of people, equipment, facilities, and money to satisfy measures of organization effectiveness and efficiency. I term these new resource combinations enabling systems in that they are necessary to move the O from an “as is” state in O-R-E-T to a “should be”’ state in O-R-E-T. Only at this point, after functional analysis and functional process description, in my judgment, can one design optimum configurations of resources and recommend to client management resource allocation alternatives in terms of numbers of people, size of budgets, amounts of inventory, and organization form and shape. This thinking appears consistent with the broader view of employing quantitative methods or opera- tions research techniques (Saaty, 1972). One may deal concurrently with pro- cess in the organization development sense, but I am convinced that the infor- mation and procedural logic must be correctly designed as a proper setting for organization development. Testing Assumptions and Crafting Recommendations Herbert Simon (1956) has pointed out: How complex or simple a structure is depends critically upon the way in which we describe it. Most of the complex structures found in the world are enormously redundant, and we can use this redundancy to simplify their description. But to use it, to achieve the simplification, we must find the right representation. And I have not found any single approach that invariably leads to the “right” representation. To test assumptions concerning the functional structure and process descriptions and to suggest feasible alternative enabling system designs, John Burns and I have adapted simulation and gaming techniques. With a ORGANIZATIONAL CHANGE . 11 general systems perspective, problem identification and solution can be de- picted as shown in Figure 2. As previously discussed, each function relates to an environment and the environment has been defined in a manner which permits comparison of the present relationship (7,) with a desired relationship (74). It should be noted that there may be two kinds of differences, with various atten- dant measures of organization effectiveness and efficiency. The desired state is different from the “‘as is’ state in kind or degree or both. In either case, the difference presents a problem and the resolution of the difference involves prob- lem solving. The information character of problem solving begins by developing for a particular o, r, and ea problem environment (PE). There may be different prob- lem environments associated with one function of O or any of its constituent os. For example, performing a printing plant scheduling function is different for job shop environments and self-contained shop environments. This example is fairly obvious, but in practice these differences are often overlooked, especially when the functions or environments are more abstract. Moos (1973) discusses important considerations in conceptualizing human environments. The prob- lem solving process begins by defining the problem environment PE—that is, by selecting appropriate segments e of the overall environment E and defining the desired relationship, r; that the organization (O) or an element of the organiza- tion (0) should have with those segments. The present relationship r, is then compared to rz. The comparison is first articulated in terms of the differences Ar. After the various factors causing the differences are determined, the various strategies sketched in Figure 2 are used. The solution must then be communi- cated to various organizational elements 0 (labeled in Figure 2 as a, b, and c) and implemented; this causes a new relationship r, to exist between O and E. That relationship must be evaluated. If it is found not adequate, the solution must be appropriately modified, which may require iteration of part of the problem solving cycle. Problem-identification or problem-solving teams of client personnel orga- nized by function and level, interdisciplinary teams of client specialists, and games or exercises involving consultants and client managers can be used any- where from the initial organization mission specification through the steps of assumption testing to the suggestion of alternative enabling system designs. For example, I have frequently encountered client “communication” subsystems which consist only of information transfer description. To recommend alterna- tive ways to improve operations optimally rather than maximally in terms of communications, consultants have to work with clients to specify both informa- tion transforms and transfers. Transforms and transfers arise on both the purely logical and the interpersonal levels of communication. Transforms are the cor- PROCTOR {+ ig= PE} . DEFINE PROBLEM ENVIRONMENT . PROBLEM {fp versus rg} IDENTIFICATION . PROBLEM DEFINITION {fa-fp =A r} . PROBLEM Ar—+O e.g. either rp is to approach p SOLUTION rg, OF rg is deemed unrealistic and a new fq is selected or rg is relaxed to approach rj , resulting in the new desired relationship Rad . SOLUTION ria COMMUNICATION ie ra . SOLUTION ae IMPLEMENTATION N . SOLUTION EVALUATION {tr-tng = A tn} Fig. 2. Problem solving process in terms of relationship, r, analysis and modification. ORGANIZATIONAL CHANGE 13 nerstone—the necessary (but not sufficient) condition of ensuring effectiveness. Transfers support transforms in the same sense that activities support functions. Transfers must be measured in terms of effectiveness in the same sense—they exist not for their own sake but only to accomplish and disseminate transforms. One method I have found workable, using the O-R-E-T model to define and develop transforms and transfers, is to combine operations research (OR) and organizational development (OD) techniques in the design, staging, and devel- opment of manual or computer-assisted simulations called controlled exercises (Proctor, 1963). This combination of techniques takes the form of personnel team building within a context of problems of a concrete nature specifically designed in such a way that the preferred problem solution is explicit. Devia- tions from interim and final problem solution steps can be observed and pro- cesses can be quantitatively analyzed and evaluated. O-R-E-T Summary This O-R-E-T model presentation is undoubtedly highly abstract and mecha- nistic; it even appears to be a static model for improving organization structure and process. I fully appreciate the inadequacies of this form of presentation. In practice much more is considered, but this simplified preparation helps me set the stage for certain specific conclusions. Clients and consultants are heading for trouble by beginning to devise the enabling systems, or use the O-R-E-T model even with this simple formulation, without a basic statement of primary mis- sion: What are we trying to do? What business are we in or would we like to be in? Without that formulation, one can’t find an E or an R or even a hint of the formulation of effectiveness and efficiency criteria. Organization improvements proceed from this more fundamental goal, whose values precede rather than follow the O-R-E-T model because those values must be articulated before one can select a specific E from the universe of potential Es. I have come to believe that these values, which many appear to assume are predictable, “given talented people using modern techniques,” are by no means certain (Beer, 1979 and 1981). The questions ““What business are we in?” or ““Who are we?” are dis- cussed and debated, resolved or left unanswered because of values, or what more modestly could be termed options. That is, what managers perceive and desire causes them to select options which precede and determine the directions taken by their systems. One could argue that these options (“opts’’) are in turn con- strained by larger systems, but a general system approach suggests that these larger systems are themselves functions of perceptions and desires which pro- mote super options (“‘super opts’’). So it seems reasonable to assume that under- 14 PROCTOR lying and directing all formal or technological systems are apparently a set of prior values or options. Opts and Super Opts Human social organizations have a “structure of mutual expectations” (Vickers, 1957), the articulation of which has been discussed here in terms of setting goals and objectives. However, there are no guarantees that these hu- mans, acting themselves as goal-directed systems within higher order systems, necessarily share a set of such expectations at the same or different levels. Nor is it sure, as this discourse may have led the reader to believe, that they interact only as formal information-processing problem-solving elements. As a practi- tioner using operations research and organization development techniques, I have observed that interactions often have a thrust of their own, independent of the institution and of their given problem-solving role within that institution. In other words, from a goal-changing point of view, we are confronted not only with various types of interpersonal interactions, but also with the possibility of self-defined and non-mutually-supportive interpersonal interactions. Concur- rence of purposes, as idiosyncratically conceived at various levels of aggregation, is not guaranteed. It is not a question of “logic”? so much as a question of premises. The articulation of these premises is crucial for those of us engaged in the design, operation, combination, and renewal of organization structures and processes. Our assumptions about why “they” behave as they do, individually or in the aggregate, are critical. The predictability of either collective or individual performance on which we all count to feed, fuel, heal, worship, and govern is profoundly impacted by the phenomenon of collective (convergent) and con- flicting (divergent) options. Weare faced, it appears, with a system of values or collection of options which produces various priority orderings of organization objectives—a system gain- ing increasing attention within and across scientific disciplines (Laszlo, 1973; Vickers, 1973). At the highest decision making level in any organization, the top manager, acting either as an individual or on behalf of the owners or power possessors, chooses an option (Lundberg, 1968), 1.e., “opts” for a specific rela- tionship between O and an E. The overall selection of goal or purposes may or may not coincide with the prevailing climate of social values at some level of ageregation. A dynamic and changing relationship between institutional goals and social values may begin. | One might argue that the environment produces constraints that drive the selection of the desired ends. In a cosmic sense, over an eternity, this may be absolutely true. In a relative moment, however, the decision maker has a consid- ORGANIZATIONAL CHANGE 15 erable apparent range of choice, within definite constraints. An individual within an organization may not opt for the success of some portion of the organization, even his own. The impact of choice which defines O, R, and E becomes generally greater as we move up what might be called the “power scale”’ toward some point which may be termed the culture itself (Boulding, 1956; Ackoff and Emery, 1972). At that level we encounter super opts. It is important to recognize their pervasive character. Serge Moscovici (1972) has addressed the importance of these options for any methodology of social understanding and has devised a provocative set of exam- ples of the radical nature of social values for research in social psychology. Studies in group dynamics are made from the viewpoint of work efficiency and increased productivity, not job satisfaction. Studies of change have been under- taken whose aim has been to reduce the resistance of one group to the goals of ‘another (labor and management respectively in the example discussed). Strate- gies of conflict resolution are based on the value of clashes between nation- states, not on ideologies or even material interests such as food. (This is a politi- cal question which assumes the values of social units.) Some economists see rate of return on investment as the fundamental value, while others select full em- ployment. In the species of goal-seeking human social interrelationships called manage- ment systems, the general system/cybernetics approach has led me to adopt a stance compatible with that expressed by J. Dennis Nolan (1974), who said, in speaking of the functional analysis of behavior, “Since functional analysis is neutral with respect to the desirability of behavior, it necessarily cannot specify the alternatives (for the products of behavior’)’. Analyzing B. F. Skinner’s approach, Nolan concluded that, rather than creating a value-free science, it represents a technological strategy that begs the question of values. In short, when we consider intentional organizational change in goal-seeking human social systems, we are confronted with the questions ““Who am I?” and ““‘Where am I going?” on a level which transcends the construction of our techno- logical systems model. And dealing on higher levels of aggregation does not provide the answers. For whatever the O, the questions remain, and furthermore exhibit themselves as a layer upon the Os. Conclusions During three decades, I have found that general systems theory and cyber- netics provides a conceptual framework or point of view which is most helpful in organizing and formalizing intentional efforts to change organizations (Proctor, 1985). There are ways of first distinguishing and then synthesizing process and structure as well as dealing with related questions of effectiveness and efficiency. 16 PROCTOR In addition, there are ways of synthesizing formal logical problem-solving and human problem-solving processes in the context of concrete problem-solving situations faced by the organizations with which we deal. Asa paradigm for a scientific, operational approach to consulting with organi- zations, general systems/cybernetics continues to clarify hypothesis formulation and evaluation capabilities. However, the necessary methodological focus on objectives, goals, and missions has led me to believe that individual and collec- tive options arising from values lie at the root of these organisms. ““While man- agement is a discipline—that is, an organized body of knowledge and as such applicable everywhere—it is also ’culture.’ It is not value free science.” (Drucker, 1974; Rifkin, 1987) These values are real. To fail to articulate both the structure and process of interacting values presages failure for consultant and client. To study them ina laboratory or controlled experiment is important. But, if utility (or “get results’’) is the name of the game, these laboratory experiments are transferable to the external world only to the extent to which they are empirically “correct” at their chosen level of human system aggregation. Geoffrey Vickers points out that “‘when we set out to shape our institutions—even to form a company—we are not creating order out of chaos. We are intervening in a dynamic situation already regulated by its own laws.” (Vickers, 1957) It seems to be exceedingly difficult to devise useful predictive models of these kinds of human-centered organizations without at a minimum explicating the ends and values involved. At least this is true for prediction of organizational effectiveness and efficiency and for classification of structure and processes as well as growth and decay. Controlled experiments involving the effects of choice are only valid if the existing world system operates in conformity with the controls placed by the experimenter. This is exactly the question which arose in management by objectives: Whose objectives? The conclusion, then, is that general systems/cybernetics can provide a useful scientific paradigm with conventional scientific methods in the sense that it proposes an interdisciplinary theoretical framework for analytical investigation of highly complex human-centered organisms. However, there are bounds. In the end, scientific investigation of human social organizations cannot be value- free. The values of any theory or experiment must be stated and only in this sense is a dialectic possible between traditional science and general systems theory, between theory and practice, between the laboratory and the practical world (Caws, 1968). Even using general systems theory, transferring models, research, or experi- ments from the laboratory to the real world is risky at best, and the subject requires extensive open debate. Models, computer-assisted or otherwise, which ORGANIZATIONAL CHANGE 17 cannot frame or account for the multiplicity of and convergence/divergence of values and their enormously complex interactions may:be misapplied. Scientists, philosophers, and practitioners constantly must take responsibility for articulating their own values, their conception of the ends and objectives of the social organisms which they perceive, theorize, or build models about and experiments upon, especially when they may possess perceptions which may be capable of directly influencing institutional values. It would be nice for all, from any vantage point, if various values were mutu- ally agreed upon so that acts achieved goals satisfying all. If not satisfying all, they would be judged to be at least minimally injurious, in some qualitative sense. Many legal, political, or moral “controls” are instituted toward that end, but the devisers and appliers of these controls are operating from a point of view. The question of whose objectives, whose values, and what values transcend what _values becomes very real to consultants, theorists, managers, and modelers working for or with companies and government agencies, or other aggregates of human social organisms. Practical improvement of certain organizations at certain times is possible. However, there are no guarantees of sustained future improvements and no guarantees that formal models can predict or formulate the nature of social organisms. In fact, scientific methodologies in this realm, to be of any lasting consequence, must articulate values and take a stand upon attendant matters of underlying ethical choice. If this is done, the world, even if never ideal, may see substantive improvement. People sense this. Hopefully, scientists and philoso- phers do too. References Ackoff, R. L., & Emery, F. E. (1972). On ideal-seeking systems. General Systems Yearbook, 17:17-24. Beck, A. C., and Helman, E. D. (Eds.) (1972). A practical approach to organization development through MBO—Selected readings. Reading, MA: Addison-Wesley. Beer, S. (1960). Below the twilight arch. General Systems Yearbook, 5:15-17. Beer, S. (1979). The heart of the enterprise. London: Wiley. Beer, S. (1981). Brain of the firm. London: Wiley. Bertalanfy, L. V. (1956). General systems theory. General Systems Yearbook, 1:3. Boulding, K. (1956). General systems theory—the skeleton of science. General Systems Yearbook, 1:11-17. Braham, M. (1973). A general theory of organization. General Systems Yearbook, 18:13-24. Caws, P. (1968). Science and system: On the unity and diversity of scientific theory. General Systems Year- book, 13:11. Drucker, P. (1974). Management tasks-responsibilities-practices. New York: Harper & Row. Fraser, J. T. (1975). Of time, passion, and knowledge. New Y ork: Georges Braziller. Hall, A. O., & Fagen, R. E. (1956). Definition of system. General Systems Yearbook, 1:21. Laszlo, E. A. (1973). A systems philosophy of human values. Behavioral Science, 18:242-249. Lundberg, F. (1968). The rich and the super rich. New York: Lyle Stuart. Medawar, P. (1973, May). What’s human about man is his technology. Smithsonian, 4:22-29. Miller, J. G. (1978). Living Systems. McGraw-Hill, New York. Moos, R. H. (1973). Conceptualizations of human environments. American Psychologist, 28:652-665S. Morris, R. (1984). Scientific attitudes toward time. New York: Simon & Schuster. 18 PROCTOR Moscovici, S. (1972). Society and theory in social psychology. In J. Israel & H. Tajfel (Eds.), The context of social psychology: A candid assessment (pp. 17-68). London: Academic Press. Nolan, J. D. (1974). Freedom and dignity: A functional’ analysis. American Psychologist, 29:157-160. Proctor, J. H. (1963). Normative exercising: An analytical and evaluative aid in system design. JEEE Transac- tions On Engineering Management, EM-10:183-192. Proctor, J. H. (1985). Speculation about pain and stress. In L. Ng & J. Proctor (Eds.), Management of pain and stress, past, present, and future. Washington: The American Division, World Academy of Art and Science. Rifkin, J. (1987). Time wars: The primary conflict in human history. New York: Henry Holt and Company. Roeber, R. J. C. (1973). The organization in a changing environment. Reading, MA: Addison-Wesley. Saaty, T. L. (1972). Operations research: Some contributions to mathematics. Science, 178:1061-1070. Simon, H. A. (1956). The architecture of complexity. General Systems Yearbook, 1:63-76. Sutherland, J. W. (1973). Axiological predicates of scientific enterprise. General Systems Yearbook, 19:3. (Quotation originally appeared in Sutherland, J. W., A general systems philosophy for the social and behav- ioral sciences. New York: Braziller.) Vickers, G. (1957). Control, stability, and choice. General Systems Yearbook, 2:1-8. Vickers, G. (1973). Motivation theory—a cybernetic contribution. Behavioral Science, 18:242-249. Weaver, W. (1948). Science and complexity. American Scientist, 36:536. Journal of the Washington Academy of Sciences, Volume 82, Number 1, Pages 19-36, March 1992 Can Hydro-Reservoirs in Tropical Moist Forest Become Environmentally Sustainable? Robert Goodland Environment Department, World Bank, Washington, DC Anastacio Juras Environment Department of Eletronorte, Brasilia Rajendra Pachauri’ Tata Research Institute, New Delhi and President, International Power Engineering Society ABSTRACT Today’s polarization of society “for” and “against” big hydroprojects relates to environ- mental costs, particularly borne by vulnerable ethnic minorities and the poor; such costs include species extinctions and tropical deforestation. This counter-productive polarization can be reconciled by transparency of planning, pluralism involving the society and especially all affected people, and by engendering national consensus on the best project. Detailed criteria for consensus are discussed. These include promotion of energy efficiency and con- servation, ranking of alternatives to the next hydro project, and environmental ranking of potential sites. Environmentally well designed hydro can be preferable to alternatives (coal, nuclear), and most environmental costs can be prevented, thus making hydro renewable and sustainable. To observers such as ourselves, society in an increasing number of tropical forest-owning countries seems to have become polarized into two ' Respectively: Environment Department of the World Bank, Washington, DC. Fax: 202-477-0565; Envi- ronment Department of Eletronorte, Brasilia, DF; Tata Research Institute, New Delhi, and President, Interna- tional Power Engineering Society. 19 20 GOODLAND, JURAS, AND PACHAURI extremes—supporters and opponents of big hydroprojects.? The media inform us about opponents brandishing machetes at confrontations with power engi- neers in the Amazon, thousands of demonstrators opposing dams in several countries, hundreds of thousands of signatures on petitions to the United Na- tions received by the Secretary-General, and even an international celebrity, Baba Amte, starving himself to death on the banks of the Narmada river in India. One government is alleged to have fallen partly because of a hydroproject proposed for a valuable southern rainforest, and several projects slated for tropi- cal forest areas have been canceled or indefinitely postponed partially or entirely on environmental grounds.? This tropical forest dam controversy transcends helping the power utilities win consensus and defuse polarization. We are concerned with global sustaina- bility.* One of the most effective ways to achieve sustainability is to accelerate the transition to renewable energy. Hydroprojects that have lower impacts and higher benefits than alternatives may play a substantial part. We postpone to another occasion the debate on large versus small projects. We are certain that the world cannot afford business as usual. This is doubly true for big dams: there is not enough capital available at affordable cost to meet projected demand for power (Imran & Barnes, 1990; Moore & Smith, 1990; World Bank, 1989). The political polarization for and against hydropower seems most extreme in coun- tries with tropical forests. This is understandable because tropical forests are often associated with major untouched rivers and are also the world’s richest source of biodiversity. Such countries exist in Latin America, Africa, and Asia. So this is very much a global debate; it is not restricted to one or two countries. Both poles could be perceived as adopting extreme positions and unwilling to explore any middle ground. The most promising approach to reconciliation is to build on the progress in hydroproject design (Table 1), broadening the constitu- * The most comprehensive documentations of this polarization are Goldsmith and Hildyard’s (1985-1991) three-volume opus, and Williams (1991). 3 Recent costly dam fights, mainly over environmental issues, include the following: India’s 240-MW Silent Valley hydroproject in Kerala’s remnant rain forest was cancelled in 1980. Thailand 1986 Nam Choan: 2000 MW was lost after feasibility stage. Thailand 1991 Pak Mun: the dam was relocated and its height lowered; delayed but now proceeding. Brazil 1988 Babaquara: 6000 MW was lost, due to campaign by rock singer Sting. India Narmada: 5 year delay after feasibility, new investigation (1991-2) awaited. Australia’s 180-MW Frank- lin River in Tasmania’s World Heritage Rain Forest was shelved in 1983. (Commissao Pro-Indio, 1989; Margulis, 1990; Paiva, 1977, 1982; Rosa, 1989; Rosa et al., 1988; Santos & de Andrade, 1988). “ Sustainability as a concept has been formally endorsed as an official priority of the United Nations system, and by the World Bank. Although the Bank knows more about the concept than it is comfortable admitting, it is difficult to operationalize the concept in all work. As they depend on the hydrological cycle, hydroprojects are theoretically renewable indefinitely. Sustainability here refers to two levels. First, the environmental and social costs—often not fully internalized—must be valued and clearly outweighed by the benefits. Second, the life of the project must not be damaged by environmental abuse, such as rapid sedimentation due to lack of watershed management upstream. Daly and Cobb (1989) have thought through the concept of sustainability the furthest. Other references include Adams (1990), Goodland et al. (1991), and Goodland & Daly (1991, 1992). HYDRO-RESERVOIRS IN TROPICAL FORESTS 21 Table 1.—Broadening the Design Constituency of Hydroprojects Design Team ' Approximate Era 1. Engineers Pre-WWII Dams 2. Engineers + Economists Post-WWII Dams 3. Engineers + Economists + EIS° Late 1970s 4. Engineers + Economists + Environmentalists Late 1980s 5. Engineers + Economists + Environmentalists + Affected People Early 1990s 6. Engineers + Economists + Environmentalists + Affected People + NGOs (Non-Governmental Organizations) Late 1990s 7. Engineers + Economists + Environmentalists + Affected People + NGOs + National Consensus Early 2000s ? Note: These dates hold more for industrial nations than for developing ones, although meaningful consulta- tions with affected people or their advocates and local NGOs, and the involvement of environmentalists in project design are now mandatory for all World Bank-assisted projects. -ency. The aim is to promote a national debate to ascertain whether there are criteria under which some acceptable and sustainable reservoirs could be devel- oped in tropical forest regions. We believe that both the transparency in decision-making and the pluralism necessary for success in such a debate will themselves significantly contribute to consensus-building. The whole process, including access to consolidated bud- gets, must be transparent so that the identities of the recipients of subsidies will be known to all. To achieve pluralism, academia, NGOs, the private sector, and the government must be included. This requires a certain amount of decentral- ization, especially of mitigatory measures. Full participation, especially of af- fected people and their advocates, also is essential. This brings responsibility: all groups must be held accountable for objective performance standards. In addi- tion, environmental standards for development projects are improving. There- fore, because a reservoir may take twenty years from investigation to comple- tion, today’s best practice is the minimum acceptable standard. Let us assume that national criteria can be agreed upon and that they can be substantially met. Are there conditions under which such a reservoir could be justified? Our opin- ion is yes. Many hydroprojects are fraught with impossible environmental prob- lems, but for others, such problems can be solved—although with much more effort than is expended today. On the other hand, hydroprojects with large reservoirs also may have major side benefits, such as flood control, improved water quality, and fisheries. Under certain conditions, recreation, tourism, irri- gation, and navigation can be made compatible with hydropower. Even without the added benefits of hydropower, the environmental problems of coal (e.g., > EIS: Environmental Impact Statement, which was started when the design was complete—a recipe for confrontation and waste. 22 GOODLAND, JURAS, AND PACHAURI Table 2.—Environmental Ranking of New Energy Sources (Simplified) BEST . SOLAR (+ hydrogen) . PHOTOVOLTAICS . WIND RENEWABLE & SUSTAINABLE . TIDAL & WAVES BIOMASS (+ alcohol) HYDRO POTENTIALLY SUSTAINABLE . GEOTHERMAL GAS OIL NON-RENEWABLE & UNSUSTAINABLE . COAL . NUCLEAR —=SOWIDNARWNH— — et WORST Note: As energy efficiency and energy conservation are becoming recognized as supply options in tropical nations, they would top this ranking. carbon dioxide, greenhouse effect) and of nuclear power® are much less solvable (Table 2). Now that economic development is being decoupled from energy consumption, damping of electricity demand need not constrain economic de- velopment for many developing countries. On the contrary, the less investment needed in energy, the more becomes available for job creation elsewhere. What might such national criteria be? Producers, consumers, and govern- ment should compile their own lists and their own ideas on the criteria they judge necessary. The present list is suggestive only. The criteria-setting process must be widely transparent in order to engender national consensus. The pur- pose of this paper is to present the case that tropical forest reservoirs meeting such national criteria could be made environmentally acceptable. Assume that a hypothetical nation, some of which is tropical forest, has reached national consensus that more energy is needed. Agreed-on criteria have been met. Efficiency and conservation have been substantially achieved. Brownouts, load shedding, and rationing are unavoidable in the near future. The choice is among coal, nuclear, and hydro. All gas and oil has been exploited or is not economic. The scope for interconnections with neighboring countries ° The nuclear industry has spent about 75% of total R & D budgets over the last four decades, but even now only generates 3% of global commercial energy. Rather than earning a profit after all these subsidies, the industry has abandoned nuclear plants, causing $10 billion in losses for shareholders in the US alone. As 10 000 to 20 000 new nuclear plants, or one new plant every three or four days, would be needed over the next 40 years to replace coal, the use of nuclear power is highly inadvisable. The enormous number of victims of the 1986 Chernobyl accident, a number likely to exceed 4 million, will postpone any recrudescence of nuclear projects. If even the skilled and disciplined Japanese can be crippled by the “very serious” 9 February 1991 accident in Mihama, the possibilities in 10 000-20 000 new plants are not reassuring. If the problem of radioactive waste storage is solved, and, in addition, if “inherently” safe designs are achieved, then prospects would improve. HYDRO-RESERVOIRS IN TROPICAL FORESTS 23 has been exploited or is not feasible. (This is important because interconnec- tions enable a more acceptable site in a neighboring country to be taken up before a worse site in the country in question. Cooperation between Uganda and Kenya is a case in point.) The crucial national question urgently needing resolution is: Is there a set of criteria that could justify reservoirs in tropical forest regions? Although this would apply to tropical forest dams in general, important country-specific crite- ria also would be necessary. Thus trade-offs are being faced in many countries, such as those between massive increases in coal-burning on the one hand and constructing dams—such as the world’s biggest dam, Three Gorges in China, and the Narmada dams in India—on the other. Global common property issues such as carbon dioxide accumulation and biodiversity conservation should not be compromised by country-specific criteria. Sector Criteria For the purposes of this discussion, we assume that the following conditions prevail: The price of electricity must substantially have already reached long-run marginal cost. Practically all consumers must have been metered; meters are substantially precise; major arrears cause prompt cessation of service. The finan- cial stability of the power utility has to be ensured. Most energy conservation and efficiency measures must substantially be in place, both in generation and transmission, as well as inside homes and factories (Shepard, 1991). That is, the marginal economic cost (including environmental externalities) of saving an additional kilowatt-hour through conservation must be as high as the marginal cost of a new one produced and delivered to the consumer. This follows from the assumption above. Because conservation mea- sures are always advancing, implementation will always lag behind savings po- tential. The goal is to minimize this lag. Also, conservation cannot reap results overnight because of restraints on the pace of replacing capital stock, plus other ‘factors. A long-term, least-cost energy services perspective is needed. And “least” cost here must fully include environmental and social costs borne in the future (inter-generational costs). Because markets are not perfectly efficient, decoupling of profits from sales is in the utilities’ interest, so that they can make money on margin and not on volume. Progressive utilities have started selling conservation to consumers. Utilities should be rewarded for efficiency. (Anan- dalingam, 1987; Chandler, 1985; Fickett et al., 1990; Flavin, 1985, 1986; Flavin & Durning, 1988; Gamba et al., 1986; Goldemburg, 1988; Guzman, 1987; Hagler, Bailley Co., 1987; Holdren, 1987; IEA, 1987, 1988; Januzzi et al., 1991; Johnson, 1989; Lovins, 1990; Moreira, 1989; Naviglio, 1990; Noguiera, 1991; Procel, 1990; Van Domelen, 1988; World Bank, 1989a,b.) 24 GOODLAND, JURAS, AND PACHAURI Discounts encouraging overconsumption by large consumers must have been repealed. However, the companies need not have been penalized to the extent that they start to generate their own power, with possibly worse impacts. Large consumers must have shifted to less electricity-intensive methods where eco- nomically feasible. (Aluminum smelting will always be energy intensive.) Na- tional energy efficiency equipment standards must be in place. Cogeneration potential must have been rationally exploited. All economically perverse subsidies and other incentives must have been rescinded. For example, some electricity and fuel pricing policies mandate that electricity and gasoline/diesel prices be the same at the power plant, refinery, port, or capital city as they are at the farthest frontier outpost. Such policies promote excessive consumption of fuel, distort industrial, population, and agri- cultural-siting policies, raise prices in the main load centers, and discourage efficient energy production in remote areas. All rehabilitation and expansion of existing sites must already have been accomplished. This is almost always achieved at much less environmental and economic cost than construction of new sites. The large number of hydropro- jects completed in the 1950s and 1960s can be modernized to postpone the need for new projects. Owen’s Falls, for example, only turbines 50% of the avail- able water. Dam Criteria As outlined above for sector criteria, we assume all reservoir sites outside tropical forests either already have been developed or are not socially, environ- mentally or economically acceptable. The following six criteria are proposed (ordering does not imply priority). First, the proposed dam should have a high ratio of power production to area inundated. Some hydroprojects have no reservoirs. On a ratio of kilowatts per hectare,’ the reservoirs with the highest ratios, in the many hundreds, include Pehuenche, Guavio and Paulo Afonso: all exceed 100.® The lowest-ratio reser- voirs include Brokopondo, Balbina, Sobradinho, Samuel, Babaquara, and Curua-Una, all under 5. Babaquara’s low ratio contributed to its cancellation. A few, such as Suriname’s Brokopondo and Burkina Faso’s Kompienga, have ratios less than 1. One admittedly arbitrary criterion or cutoff point could be 30, ’ This is a serviceable but arbitrary ratio. Both GWh and Kwh/ha would be better indicators. 8 This paper focuses on hydro-reservoirs, which are increasingly common in tropical moist forest, rather than on irrigation reservoirs, which do not occur in tropical moist forest; some multipurpose reservoirs and those in tropical dry forest remnants (e.g., India’s Narmada) are mentioned. Irrigation reservoirs may be slated for the tropical dry forest remnants; these would be even more problematic than those in tropical wet forest. HYDRO-RESERVOIRS IN TROPICAL FORESTS. 25 as in Tucurui (Table 3). Clearly, this depends on an expanded cost-benefit analysis in each case. If the ecosystem to be flooded is intact primary tropical forest, the ratio should be set much higher, say 100; if the ecosystem is agricul- tural or degraded land, then the ratio should be set lower. Economists are strug- gling to assign prices to intangibles, irreversibles, and intergenerational equity, such as those involved in the extinction of species. Second, the proposed site and surroundings should have had a thorough biotic inventory, and there should be no centers of species endemism, rich biodiversity, or other special features. The ecosystem of the proposed site should be very well conserved in perpetuity nearby, as a compensatory area ecologically equivalent to or better than the flooded area. The biotic salvage will be effective.’ Third, the reservoir water retention time should be brief—days or weeks, rather than many months (1.e., a rapid circulation rate). The shorter the reten- tion time, the less time there will be for anaerobic conditions to be created, and the better will be the water quality in the impounded area, as well as downstream for all uses.'° The nearer to “run-of-river” the project is, the fewer will be the environmental problems. The tradeoff here will be between valuable inter-sea- sonal and over-year regulation, which can be less necessary in seasonless rain forest areas. Less regulatory capacity means fewer benefits from flood control. This essen- tially means two types of sites are especially valuable. The first type is a canyon in which the reservoir does not rise above the top; these do not need large flows. Harnessing waterfalls that fish never ascend prevents migratory fish problems. The second type is no-head in-stream axial turbines which do not flood any forest. The best sites have such low volumes of biomass that decay will neither contribute significantly to greenhouse gases, nor impair fish and water quality, nor waste valuable biomass, nor clog turbine intakes. Removal of economically extractable biomass decreases greenhouse gas production and water quality risks. By inventing submersible chainsaws, Eletronorte (the federal power “agency for the Amazon region of Brazil) utilizes already inundated trees in their Tucurui reservoir (Cadman, 1991). Brief retention time has to be balanced with storage needed for irrigation and navigation. Fourth, there should be no vulnerable ethnic minorities living in or using the general area of the proposed site. No other settlements should be affected, unless * Live rescue for release into biotically impoverished habitat or into zoos and arboreta has rarely been effective historically, although captive breeding and reintroduction merits invigoration. More cataloguing and preservation of seeds and dead specimens also is urgently needed. '© Decaying tropical forest generates massive volumes of greenhouse gases, especially methane, which is 32 times more damaging than carbon dioxide. Large, shallow reservoirs from which forest is not removed may generate vastly more greenhouse gas than a coal-fired thermal equivalent (Gupta & Pachauri, 1990). 26 GOODLAND, JURAS, AND PACHAURI the livelihood of the deported after resettlement is guaranteed to be better than it was before, as measured by systematic socio-economic surveys. Higher firm- Gwh/family-displaced ratio projects should have preference. But more signifi- cant is the subsequent improvement in livelihood. This means the proposed site has been thoroughly assessed by sociological and anthropological professionals, well before any decisions have been made. Direct internalization of costs needed for adequate resettlement may be acceptable for normal deportees. But for vulnerable ethnic minorities, experience shows that it has not yet been possible to achieve adequate resettlement. If roads, employee housing, and construction materials are available near the proposed site, that will reduce the impact further. Sometimes it is not obvious who the beneficiaries are. For example, in tropi- cal forest reservoirs used for export aluminum smelting, the beneficiaries are the industrial countries’ aluminum consumers. The question then arises: For whom are the tropical hydroproject owners deciding? Similarly, in James Bay, the Cree claim the land flooded; the Quebec and Canadian governments also have claims. Is North America the beneficiary? These questions highlight the need to address explicitly the tradeoff between the beneficiaries and the people bearing the costs. Fifth, there should be no water-related diseases, such as malaria, Japanese ’B’ encephalitis, or schistosomiasis anywhere in the general region. Nor may they be likely to arrive. The risk of their arrival is reduced by destruction of the nearest foci. If water-related diseases are present, they must be eradicated, preferably before the impoundment creates more habitat. If this is impossible, the diseases should be controlled to the best extent feasible and a public health component should be integrated into project design. Sixth, the proposed dam should be sited above undammed tributaries, to help minimize changes in flood regime (on which wetlands depend) and to provide alternative upriver sites for migratory fish. There is much uncertainty even in the relatively very simple, depauperate Northern fish biological systems and in their behavior related to impoundments. Certainly much more effort than at present is needed to increase the benefits and opportunities from fisheries. Also, the dam should be proposed for an already dammed river. From the environmental point of view, dams should be concentrated on already dammed rivers, rather than siting one or a few dams on a larger number of rivers. Thus, a representative sample of the nation’s rivers would remain in their natural, free- flowing state. This tradeoff with the risk of low flows curtailing power output should not be common to the extent tropical wet forested catchments are not usually seasonal. In multi-purpose dams, the enormous value of the annual flood restoring productivity downstream should be factored in. HYDRO-RESERVOIRS IN TROPICAL FORESTS 27 Table 3.—Hydropower Generated per Hectare Inundated (Examples only) Final Rated Normal Area of Capacity Reservoir Kilowatts Project (Country) (MW) (ha) per Hectare Paulo Afonso (Brazil) I-IV 3984 1600 2490 Pehuenche (Chile) 500 400 1250 Guavio (Colombia) 1600 1500 1067 Rio Grande II (Colombia) 324 1100 295 Itaipu (Brazil and Paraguay) 12600 135000 93 Aguamilpa (Mexico) 960 12000 80 Sayanskaya (USSR) 6400 80000 80 Churchill Falls (Canada) 5225 66500 79 Grand Coulee (USA) 2025 32400 63 Urra I (Colombia) 340 6200 5 Jupia (Brazil) 1400 33300 42 Sao Simao (Brazil) 2680 66000 41 Tucurui (Brazil) 7600 243000 Si Ilha Solteira (Brazil) 3200 120000 25) _Guri (Venezuela) 6000 328000 18 Paredao (Brazil) 40 2300 17 Urra II (Colombia) 860 54000 16 Cabora Bassa (Mozambique) 4000 380000 14 Three Gorges (PRC) 13000 110000 12 Furnas (Brazil) 1216 144000 8 Aswan High Dam (Egypt) 2100 40000 5 Curua-Una (Brazil) 40 8600 5 Samuel (Brazil) 2G, 57900 4 Tres Maria (Brazil) 400 105200 4 Kariba (Zimbabwe/Zambia) 1500 510000 3 Petit-Saut (French Guiana) 87 31000 2.8 Sobradinho (Brazil) 1050 421400 2 Balbina (Brazil) 250 236000 Babaquara (Brazil) 6600 600000 l Akosombo (Ghana) 833 848200 0.9 Kompienga (Burkina Faso) 14 20000 0.7 Brokopondo (Suriname) 30 150000 0.2 Note: This table is only partially indicative, since it does not reflect the value of the land inundated, which can vary significantly. Some of the land inundated is river bed. The more reliable ratio kwh/ha (instead of kw/ha) is being calculated. The ranking would be improved, but little altered, if river bed or normal annual flood areas were subtracted. Islands in the reservoir also could be subtracted in certain cases. Some of these figures are for non-forest reservoirs and most are hydropower, rather than irrigation reservoirs. Area inundated is the key issue. Less seasonal tropical wet forest reservoirs do not need to be large. Optimizing the tradeoffs at the margin of reservoir capacity is more important than choosing between having or not having a reservoir. It is relatively easy to include bottom sluices at the design stage for such releases. Dams should not cause species extinctions, including those of migra- tory fish that would be denied access to breeding or feeding sites. This means the damming of the last few free-flowing rivers in a region will be even more difficult to justify. In tropical forest areas, roads built to facilitate hydroproject construction or operation can “open up” significant areas to colonization and deforestation. Therefore, care must be taken during road planning and operation to reduce this eflect: 28 GOODLAND, JURAS, AND PACHAURI Criteria on Small-Scale and Unconventional Power Potentials Small-scale and unconventional potential alternatives are being examined or are already substantially exploited. Privately owned renewable energy genera- tors sell surplus to utilities. Such alternatives include: a) No-dam (or very low head) axial tube turbines within the river. b) Small generating systems (including water wheels). c) Solar power elsewhere in the country (including photovoltaics, tidal, wind and hy- drogen from splitting water molecules).!! d) Biomass energy production (biomass plantations, garbage and sewage). Many tropical forest countries contain dry, sunny, or even desertic regions where solar powered electric plants can be sited. They occupy 5% to 10% of the land of even the “best” (i.e., high head / low area) hydroschemes, and often can put otherwise unproductive land to sustainable use. We believe, and the World Bank is in process of calculating, that solar power is already economic in com- parisons with hydro when the value of inundated forest is internalized, even imperfectly. Population Stability Criteria Human population stability is an essential precondition for all sustainable use of renewable resources, including use both of hydropower and of tropical for- ests. Human populations of tropical moist forest-owning countries annually increase by more than 2.4%, which means a doubling in 25 years. Sustainability criteria will be difficult enough to fulfill without having to double the electricity supply every 25 years. The situation is more severe in those countries in which the per capita electricity use also is rising. Average planned power demand growth is about 7% in developing countries—a doubling every ten years. (In Brazil per capita use is projected to rise 55% by 2000). It is sensible to permit electricity companies to profit from their customers’ investments in conservation. Utilities should not be penalized for investments in conservation. An increasing number of Northern utilities now find it more economic, rather than to generate more electricity, to provide free fluorescent light fixtures and to promote or even to subsidize other more efficient appliances for consumers. This suggests, as we noted above, that the pricing policy is wrong in these cases. Although this requires sensible action on pricing in the power market which does not yet exist in most developing countries, the preference is '' Hydrogen from splitting water molecules is likely to become economically and technically feasible very soon (Ogden & Williams, 1989). While it is difficult to generate SOOMW from garbage and sewage now, a large number of smaller such plants reduces the need for large projects. ! HYDRO-RESERVOIRS IN TROPICAL FORESTS 29 clear. Utilities now conduct free energy audits for consumers, showing where energy can be conserved most. To the extent this holds for population, power doinondtiolis support of gov- ernmental family planning goals will reduce the national controversies and project delays commonly experienced. Power corporations already help to the extent that televisions are contraceptive. We do not want to burden the power sector with nghting all societal ills. However, population stability is so impor- tant for sustainability that family planning or similar activities should be compo- nents of all relevant projects, including those in the power sector. Case Example of Brazil The above suggestions are generic rather than specific to any particular coun- try. However, Brazil, Eletrobras (its federal power agency), and Eletronorte in - particular are deeply concerned with both energy conservation and environmen- tal impacts. So is the citizenry—if not more so (Eletrobras, 1986, 1987, 1990; Goldemburg, 1987; Holtz, 1989; Juras, 1990, 1991; Lacerda, 1990; Serra, 1991; Zatz, 1990). Brazil has probably saved more than US $1 000 000 000 in new generation capacity avoided because of recent major improvements in the elec- tricity tariff structure, which led to more conservation and efficiency (Geller, 1986, 1988, 1990). The World Bank has commended Brazil for moving towards a more appropriate tariff structure and in the direction of the difficult goal of raising the price of electricity towards the long-run marginal cost of production. The World Bank values the partnership with Eletrobras and has assisted in financing the federal electricity conservation program, under the direction of the Science and Technology Secretariat. The World Bank also is glad to be partners with the National Environmental Secretariat in the first and biggest loan solely for national environmental priorities and institutional strengthening (US $117 000 000 in February 1990). The government, Eletrobras, Eletronorte, and environmentalists are adopting a new position on new Amazonian hydroprojects as a result of evolution of environmental awareness, specific legislation, and experience with Amazonian issues (Adam, 1988; Eletrobras, 1986, 1987, 1989, 1990; Goldemberg & Bar- bosa, 1989). Current construction rankings suggest that environmental criteria are most effective when applied proactively. This emphasis on environment, conservation, and efficiency is exceptionally well placed. The recent hiring of substantial numbers of environmental professional staff by all Eletrobras’ con- cessionaires is encouraging. For example, Eletronorte’s environmental staff rose from less than one at the time Tucurui was designed in the late 1970’s to over 100 today (Goodland, 1978, 1990, 1991). Environmental training throughout the entire power sector has increased dramatically. 30 GOODLAND, JURAS, AND PACHAURI Capital availability is a major constraint on the power sector, which has been responsible for as much as 25% (US $30 000 000 000, 1973-83: now about 19%) of Brazil’s foreign debt. Eletrobras may require of the order of US $7 500 000 000 to meet its 1991-2000 demand projections. In today’s era of severely limited capital, such huge public investments in any sector, such as power supply, could force reductions in investments in other sectors, especially environment and the social sectors—education, nutrition, and health—as well as in poverty alleviation. Thus electricity, formerly a driving force behind social and economic development, could instead hinder vital welfare gains if improved pricing, conservation, and environmental precautions are not achieved. We could be entering an era in which power investments reduce investments in other sectors whose growth was the driving force underlying electricity demand projections. Electricity rationing started during the 1985-86 Northeast drought and is projected to increase in the mid-1990s. Eletrobras projects that electricity de- mand will double between 1988 and 2000. This means that 37 000 MW needs to be installed by 2000. How can we best install the equivalent of three new Itaipus —the world’s largest hydroproject—in this decade? How can we avoid repeating delays, confrontations, and wastage? Reports are guardedly encouraging. One of the next Amazonian dams may be the 1328-MW Serra Quebrada project just upstream from Tucurui. This meets many of the criteria listed above, and contrasts starkly with the Balbina/Baba- quara-type (Cummings, 1991; Dwyer, 1990; Fearnside, 1989, 1990; Gribel, 1990; Hecht and Cockburn, 1990; Moreira, 1987; Sao Paulo Energia, 1988; Visao, 1985). According to Eletronorte, there are no Amerindian settlements and little involuntary resettlement. The reservoir is small and practically run-of- river, and has a high ratio (31.5) of kW/ha of land flooded, which is slightly better than Brazil’s biggest hydroproject, Tucurui. In addition, it is on the al- ready dammed Tocantins river, rather than being the first on a hitherto un- dammed Amazonian river. _ This presages well for the ranking of the next Amazonian dams potentially identified by Eletrobras’ Plano 2010 for the next twenty years. The range be- tween the best and worst hydro sites is so wide that the least cost (after conserva- tion) power investment program will include a full array of sources, such as gas, and imported power. Coal and (possibly at some time in the future) even nuclear (with best technology) may be found by Brazil to be better than the worst potential hydro site on future ranking on national criteria. A mixed hydro-ther- mal system implies fewer reservoirs. Eletronorte has a massive challenge. Recent developments (PROCEL, can- cellation of Babaquara, criteria of Serra Quebrada) suggest promising improve- HYDRO-RESERVOIRS IN TROPICAL FORESTS 31 ments. National consensus on the kind of criteria suggested above will ensure that the trend is strongly positive. The World Bank wants to support this trend to the fullest extent possible. Case Example of India The case of India differs substantially from that of Brazil in the sense that India has not invested a substantial share of its power sector resources in hydro- electric plants. The main reasons for this are: first, that India has 148 600 000 000 tons of non-coking coal, and second, that development strate- gies have relied only to a very small extent on foreign borrowings. Even though the Indian economy has generally recorded a savings rate of over 20%, resource mobilization in the power sector remains severely constrained. _ This has happened for three main reasons. First, power sector demand growth in recent years has been rather high (9-10% annually), with a growing peak demand relative to base load demand. Second, the electric utility industry has accumulated heavy losses on account of suppressed tariffs and operational inef- ficiencies. Third, high human population growth (1.8% annually) continues to impose Onerous demands on investments in education, health care, welfare programs, and infrastructure. The power sector is thus one of many sectors competing for limited resources. These factors have resulted in a preference for relatively short gestation ther- mal power plants rather than hydroelectric capacity. While hydro and thermal had almost the same share of power generation (45% and 55%, respectively) in 1965-66, the distribution is now 30% hydro to 70% thermal. Fortunately, In- dian coal is low in sulfur, even though its ash content exceeds 40% in some power stations. As a result, the main environmental problems of thermal power stations are particulate emissions and ash disposal. Except in regions like the Rihand reservoir—now well known for the Singrauli thermal power plants— acid rain is not now a problem, nor is it likely to become much of one in the future. India’s main hydrosites are in the Himalayas, with a large share concentrated in the North-East. India has a land to population ratio of 0.004 km? per capita while Brazil’s is 0.070 km? per capita. High population densities, land scarcity (particularly agricultural), and disappearing forests are three crucial factors in Indian hydro planning. For example, the major issue in the 1200-MW (US $1 130 000 000) Sardar Sarovar hydro and irrigation project on the Narmada river is the involuntary resettlement of people. These 90,000 deportees are not well equipped to adapt to new habitats, having a historically long intergenera- tional dependence on the land and its specific biota. In addition, the track record 32 GOODLAND, JURAS, AND PACHAURI Table 4.—Indian Energy Tariffs Energy Source Economic Price Market Price Domestic Cooking (Rs./Mcal) Kerosene 7.78 305 Electricity 20.43 8.49 Irrigation (Rs./ML of water) HS Diesel 109.64 142.02 Electricity 107.19 9.03 [Note: Rs. = Indian Rupees; HS Diesel = High Speed Diesel; Mcal = megacalorie; Pachauri 1991 pers. comm.| of Indian involuntary resettlement is poor, so that hydroprojects are likely to run into heavy public resistance (Goodland, 1985, 1989; Pachauri, 1990a,b,c). Capital constraints in the Indian economy are intensifying. Therefore the impact on the power sector is likely to become more serious in coming years. Typically, the power sector has accounted for less than 20% of planned public sector investments, but the targets for the current (eighth) five-year plan demand a higher share. Therefore, energy eficiency improvements become more urgent. Conservation is important here not only at the end-use level, but also in the energy supply industry itself. For instance, official transmission and distribution losses have risen to 22%, and are as high as 40% in some states. Similarly, coal thermal efficiencies are well below state-of-the-art levels, with some plants at- taining only 20%. There are, therefore, tremendous opportunities for efficiency improvements in the power industry which would moderate new capacity growth without sacrificing electricity supply. Irrationally low energy tariffs, far below long-run marginal costs, are the main reason for lack of energy conservation. This is particularly true in the power sector wherein some end-user subsidies are extremely high. (See Table 4.) Efh- ciency improvements must begin with adjustment of energy tariffs, in order to provide the consumer with appropriate signals. Improved efficiency may not significantly reduce available aggregate demand for power to the extent that overall quantity of power also constrains. Investments in physical capital have to be matched with investments in human capital, especially in power sector planning and environmental assessment. Such human capital investments would have larger returns than almost any other form of investment in the power sector. Conclusion Our conclusion devolves on the likelihood of a country fulfilling most of the above criteria. These criteria are stringent even for industrial countries. To what HYDRO-RESERVOIRS IN TROPICAL FORESTS 33 extent will such criteria be fulfilled? We agree with skeptics who rightly claim not all these criteria will be fully met. But the process of agreeing and approach- ing the criteria will be salutary. Do sites fulfilling most criteria exist? The least bad site certainly exists. The decisions will be difficult in some cases, less so in others. Although difficult, this course 1s better than the alternatives, and much easier than damping demand until solar/hydrogen energy becomes feasible in the next decades. Mandatory rationing and other service interruptions are likely to be exceedingly painful to consumers and to the development of the country. The damping pain should be thought through and discussed with all interested parties, as part of the criteria-setting and consensus-building exercise. Proper pricing makes the choices more obvious. In sum, we need to compare costs and benefits much more rigorously and comprehensively than has been the case so far. ‘In our imperfect world, the reality is that not all these criteria will ever be totally met. Therefore, a national consensus is needed on whether or not conser- vation, efficiency, environmental precautions, and other alternatives are being pursued adequately. The national consensus 1s essential in order to agree on the threshold at which the second best—or least bad—site should be developed. National agreement on criteria will reveal where the thresholds lie. As soon as the various environmental impacts can be evaluated, the polarization of society will be defused. The need is to make uncertainty transparent and positive, rather than covert and manipulative. As hydropower is exceptionally capital intensive and capital availability is a major constraint in nearly all nations, tropically forested or not, it is imperative to follow the least-cost (as defined above to include social and environmental costs) sequence of development. Of course, least cost specifically includes saving kilowatt-hours, not just generating them, based on consumer choice when fac- ing appropriate prices. We urge the use of proper opportunity cost of capital. Arguments for simultaneous development of higher cost alternatives should be rigorously resisted. Power corporations are commendably making the transition away from sole focus on new capacity and towards conservation and efficiency. This is difficult for them because new capacity is under their almost total con- trol, whereas conservation means they have to persuade other sectors outside their control. Power corporations wanting to promote sustainability and to reduce national controversies and delays should follow a vigorous action program with serial steps along the following lines (Goodland, 1988, 1990a,b,c): 1. Promote fulfillment of agreed-on criteria. 2. Manage demand to the fullest extent justifiable. 3. Promote agreed-on valuation of impacts. 34 GOODLAND, JURAS, AND PACHAURI 4. Seek sites fulfilling nationally-agreed-on criteria. 5. Sequence all sites in a national least-cost power program, under credible scenarios. 6. Rank all potential sites on the basis of these criteria. 7. Only then, develop the least bad new site fulfilling such criteria. Acknowledgements In addition to the comments by World Bank colleagues, we want to acknowl- edge with great pleasure the support of the former president of Eletronorte (now Secretary of National Energy) Armando Araujo, David Houghton, and Howard Geller for their most helpful comments on earlier drafts of this paper. References Adam, J. A. (1988). Extracting power from the Amazon. JEEE Spectrum, 25:34-38. Adams, W. M. (1990). Green development: Environment and sustainability in the Third World. London: Routledge. Anandalingam, G. (1987). The economics of industrial energy conservation in developing countries. In Pa- chauni, R.K. (Ed.), Global energy interactions (pp. 643-661). Riverdale, MD: Riverdale Press. Balbina: economia do petroleo na Amazonia. (1986, July 16) Visao, pp. 30-33. Cadman, J. D. (1989). Energy from the Amazon. Civil Engineering, 59:55-57. Cadman, J. D. (1991). Exploracao da madeira submersa na UHE Tucurui. Brasilia: Eletronorte, Depto. de Meio Ambiente. Chandler, W. U. (1985). Energy productivity: key to environmental protection and economic progress (Paper 63). Washington: Worldwatch Institute. Comissao Pro-Indio de Sao Paulo (ca. 1989). O que e o aproveitamento hidreletrico de Cachoeira Porteira? Sao Paulo: CPI. ; Cummings, B. J. (1991). Dam the rivers, damn the people: Development and resistance in Amazonian Brazil. London: Earthscan. Daly, H. E., & Cobb, J. (1989). For the common good: Redirecting the economy towards community, the environment and a sustainable future. Boston, Beacon Press. Dwyer, A. (1990). Into the Amazon: The struggle for the rain forest. San Francisco: Sierra Club Books. Eletrobras. (1986). Manual de estudos de efeitos ambientais dos sistemas eletricos. Rio de Janeiro: Author. Eletrobras. (1987). Plano nacional de energia eletrica 2010: Relatorio executivo. Rio de Janeiro: Author. Eletrobras. (1989). Plano decenal 1990-1999. Rio de Janeiro: Author. Eletrobras. (1990). Plano diretor de meio ambiente do setor eletrico 1990-1992. Rio de Janeiro: Author. Fearnside, P. M. (1989). Brazil’s Balbina dam: Environment versus the legacy of the Pharaohs in Amazonia. Environmental Management, 13:401-423. Fearnside, P. M. (1990). Balbina: licoes tragicas na Amazonia. Ciencia Hoje, 11:34—40. Fearnside, P. M. (1990). A hidreletrica de Balbina: o faraonismo irreversivel versus 0 meio ambiente na Amazonia. Sao Paulo: Instituto de Antropologia e Meio Ambiente (IAMA). Fickett, A. P., Gellings, C. W., & Lovins, A. B. (1990, September). Efficient use of electricity. Scientific American, pp. 65-74. Flavin, C. (1984). Electricity’s future: The shift to efficiency and small scale power (Paper 61). Washington: Worldwatch Institute. Flavin, C. (1986). Electricity for a developing world (Paper 70). Washington: Worldwatch Institute. Flavin, C., & Durning, A. B. (1988). Building on success: the age of energy efficiency (Paper 82). Washington: Worldwatch Institute. Gadgil, A. J., & Januzzi, G. De M. (1991, June). Conservation potential of compact fluorescent light in India and Brazil. Energy Policy, pp. 449-463. Gamba, J., Caplin, D., & Muckhluyse, J. (1986). Industrial rationalization in developing countries. Baltimore: Johns Hopkins Press. Geller, H.S. (1986). End-use electricity conservation: Options for developing countries (Energy Dept. Paper 32) Washington: World Bank. Geller, H. S. (1988). Electricity conservation in Brazil: potential and progress. Energy, 13:469-483. Geller, H. S. (1990). Electricity conservation in Brazil. Washington: American Council for an Energy-Efficient Economy. HYDRO-RESERVOIRS IN TROPICAL FORESTS 35 Goldemberg, J., & Barbosa, M. N. (Eds.) (1989). Amazonia: Facts, problems and solutions. Sao Paulo: Universidade de Sao Paulo. Goldemberg, J. et al. (1987). Energy for development. Washington: World Resources Institute. Goldemberg, J., Johansson, T. B., Reddy, A. K., & Williams, R. H. (1988). Energy for a sustainable world. New Delhi: Wiley. Goldsmith, E., & Hildyard, N. (1985-1991). The social and environmental effects of large dams (Vols. 1-3). Wadebridge, UK: Ecological Centre. Goodland, R. (1978). Environmental assessment of the Tucurui hydroproject, Rio Tocantins, Amazonia. Brasi- lia: Eletronorte. Goodland, R. (1982). Tribal peoples and economic development: Human ecological considerations. Washing- ton: The World Bank. Goodland, R. (1985). Environmental aspects of hydroelectric power and water storage projects. Roorkee U.P., UNESCO/UNEP International Conference “Environmental Aspects of Water Projects” III: 1-30. Goodland, R. (1986, November). Hydro and the environment: evaluating the tradeoffs. Water Power and Dam Construction, 25-31. Goodland, R. (1989). Environmental aspects of hydroprojects. Indian J. Public Administration, 35:607-633. Goodland, R. (1990a). The World Bank’s new environmental policy for hydroprojects. International Environ- mental Affairs, 2:109-129. Goodland, R. (1990b). BIRD: exigencias para preservar a vida. Sao Paulo Energia, 7:20-24. Goodland, R. (1990c). The World Bank’s new environmental policy for dams and reservoirs. Water Resources Development, 6:226-239. Goodland, R. (1991). Politica ambiental do BIRD sobre hidreletricas. Sao Paulo, Universidade Paulista (Mes- quita, M. e Di Genio eds.) (no prelo). Goodland, R., Daly, H., & El Serafy, S. (Eds.) (1991). Environmentally sustainable economic development: Building on Brundtland (Environment Department Working Paper 46). Washington: The World Bank. Goodland, R. & Daly, H. (1991). Environmental sustainability in the World Bank. International Journal of Sustainable Development, 1:12-23. Goodland, R. & Daly, H. (1992). Approaching global environmental sustainability. Rome: Society for Interna- tional Development, 2:35-41; 3:64-71. Gribel, R. (1990). The Balbina disaster: The need to ask why? The Ecologist, 20:133-135. Gupta, S., & Pachauri, R. K. (Eds.) (1990). Global warming and climate change: Perspectives from developing countries. New Delhi: Tata Energy Research Institute. Guzman, O. (1987). Energy efficiency and conservation in Mexico. Boulder, CO: Westview. Hagler, Bailly Co. (1987). Financing energy conservation in developing countries. Washington: Author. Hart, C. (1991, March). A Brazilian tribe escapes extinction. World Development (UNDP), pp. 15-18. Hecht, S., & Cockburn, A. (1990). The fate of the forest: Developers, destroyers, and defenders of the Amazon. London: Verso. Holdren, J. P. (1987). Global environmental issues related to energy supply: the environmental case for increased efficiency of energy use. Energy, 12:975-992. Holtz, A. C. T. (1989). Electric generation, resources and current development patterns: The case of the Amazon. In Amazonia: Facts, Problems, and Solutions: Annals I (pp. 139-172). Sao Paulo: Universidade de Sao Paulo/INPE. International Energy Agency. (1987). Energy conservation in IEA countries. Paris: Author. International Energy Agency. (1989). Electricity end-use efficiency. Paris: Author. Imran, M., & Barnes, P. (1990). Energy demand in the developing countries: Prospects for the future (Commod- ity Working Paper 23). Washington: World Bank. Januzzi, G. de M. (1990). Conservacao de energia, meio amb:ente e desenvolvimento. Ciencia Hoje, 11:16- DD: Januzzi, G. de M., Gadgil, A., Geller, H., & Sastry, M. A. (1991). Energy-efficient lighting in Brazil and India: potential and issues of technology diffusion. Unpublished manuscript. Johansson, T. B., Bodlund, B., & Williams, R. H. (1989). Electricity: Efficient end-use and new generation technologies, and their planning implications. Lund: Lund University Press. Juras, A. A. (1990). Environmental policies as applied to Eletronorte, the Brazilian northern electrical author- ity’s enterprises in the Amazon region. Brasilia: Eletronorte. Juras, A. A. (1991). Hydropower plants in the Amazonian region: opportunities for acquiring knowledge, education and training in the field of environment (Chihuahua, Mexico, International Symposium on Hy- drology and Water Resource Education and Training). Brasilia: Eletronorte. Lacerda Neto, Z. F. (1990). Environmental impact assessment in Brazil. Environmental Education and Infor- mation, &.,. Lovins, A. B. (1990). Four revolutions in electric efficiency. Contemporary Policy Issues, 8:122-141. Movimento de Apoio a Resistencia Waimiri-Atroari. (1987). Balbina: catastrofe e destruicao na Amazonia. Manaus: Author. 36 GOODLAND, JURAS, AND PACHAURI Margulis, S. (1990). O desempenho do Governo Brasileiro e Banco Mundial com relacao a questao ambiental em projetos co-financiados pelo Banco. Rio de Janeiro: IPEA. Moore, E., & Smith, G. (1990). Capital expenditures for electric power in the developing countries in the 1990s (Industry and Energy Paper 21). Washington: World Bank. Moreira, J. R. (1989). Electricity in Brazil: conservation and other major problems of developing countries. Sao Paulo: Univ. Sao Paulo, Inst. Eletrotecnologia e Energia. Naviglio, A. (1990). Energy for development: energy conservation in developing countries. Applied Energy, 36:143-157. Noguiera, U. (1991). Can environmental concerns become a constraint on energy development? (Socio-Eco- nomic Report 1990). Washington: InterAmerican Development Bank. Ogden, J. M., & Williams, R. H. (1989). Solar hydrogen: moving beyond fossil fuels. Washington: World Resources Institute. Pachauri, R. K. (1990a). Emissions of greenhouse gases in developing countries: abatement strategies and perspectives (PRE Energy Dept., working paper). Washington: World Bank. Pachauri, R. K. (1990b). Energy efficiency and conservation in India. Industry and Environment, 13:9-24. Pachauri, R. K. (1990c). Energy efficiency in developing countries: Policy options and the poverty dilemma. Natural Resources Forum, 14:319-325. Paiva, M. P. (1977). The environmental impact of man-made lakes in the Amazon region of Brazil. Rio de Janeiro: Eletrobras, Diretoria de Coordenacao. Paiva, M. P. (1982). As grandes represas do Brasil. Brasilia: Editerra. Philips, M. (1991). The least-cost energy path for developing countries: Energy efficient investments for the Multilateral Development Banks. Washington: International Institute for Energy Conservation. PROCEL. (1990). Realizacoes e metas de conservacao de energia. Rio de Janeiro: Eletrobras, Programa Nacional de Conservacao de Energia Eletrica. Rosa, L. P. (1989). Hidreletricas e meio ambiente na Amazonia: analise critica do Plano 2010. Rev. Bras. Energia, 1:7-24. Rosa, L. P., Sigaud, L., & Mielnik, O. (Eds.) (1988). Impactos de grandes projetos hidreletricos e nucleares. Sao Paulo: Marco Zero. Santos, L. A. O., & de Andrade, L. M. M. (1988). As hidreletricas (do Rio Xingu) e os povos indigenas. Sao Paulo: Comissao Pro-Indio de Sao Paulo. Serra, M. T. F. (in press). Resettlement planning in the Brazilian power sector: Recent changes in ap- proach.(ms.34+p.) In Cernea, M., & Guggenheim, S. (Eds.), Anthropological approaches to involuntary resettlement: Policy, practice and theory. Boulder, CO: Westview. Sao Paulo Energia. (1988). Balbina: obra discutivel mas agora irreversivel. Sao Paulo Energia, 5:33-36. Scaramucci, J. A., Januzzi, G. de M., & de Castro, S. A. (1990). Environmental constraints on economic growth: A global analysis of two interacting regions, Brazil and OECD. Campinas: S.P. Unicamp. Shepard, M. (1991). How to improve energy efficiency. Jssues, 7:85-91. Van Domelen, J. (1988). Power to spare: The World Bank and electricity conservation. Washington: World Wildlife Fund. Williams, P. B. (1991, August). The debate over large dams. Civil Engineering, pp. 42-48. World Bank. (1989). The future role of hydropower in developing countries (Industry & Energy Paper 15). Washington DC: Author. World Bank. (1989). Energy efficiency strategy for developing countnes. Washington DC: Author. Zatz, J. (1990). Programmes de maitrise de l’energie au Bresil. UNEP Industry and Environment, 13:12-16. Journal of the Washington Academy of Sciences, Volume 82, Number |, Pages 37-51, March 1992 Ten Reasons Why Northern Income Growth is Not the Solution to Southern Poverty Robert Goodland and Herman Daly! Environment Department, The World Bank, Washington DC 20433 USA “Those who make $200 a year should not pay so that those who make $10,000 a year can breathe clean air.. . . We are all in the same planetary boat. A few of us travel first class, while most are in steerage. But if the boat sinks we all drown together.” Ambassador Edward Kufuor, Chairman, Group of 77, at the UN Assembly, 1991. ABSTRACT Decreasing Southern poverty is arguably today’s main goal of economic development. The two main views on how this can be achieved are not fully compatible. The traditional view, is that rich Northern high-consumption societies should consume yet more in order to help the South by providing larger markets. This paper outlines the alternative view: that the North should stabilize its resource consumption, and reduce its damage to global life-sup- port systems. Any higher consumption must come from productivity improvements, rather from increased throughput growth. If natural resources were infinite, then growth would be unreservedly good. Since resources are finite, then more Northern growth inevitably means less room for Southern growth. Productivity improvements must replace throughput growth as the path of progress for the North, and eventually for the South as well. Divergent Views on How to Reduce Southern Poverty Decreasing Southern poverty 1s arguably today’s main goal of economic devel- opment. The two predominant views on how this can be achieved are not fully compatible. The traditional view, held by most economists and development agencies, is not working well. The traditional view is that rich Northern high- ' Respectively Adviser and Senior Economist. Fax: 202/477-0565 37 38 GOODLAND AND DALY consumption societies should consume yet more in order to help the South by providing larger markets. For example, the Bretton Woods Institutions were in part created because of macroeconomic market failure, both to maintain full employment and to bridge the income gap between rich and poor, and to intermediate between rich coun- tries and poor. Their leadership role, apart from lending,” is as purveyors of ideas, as well as of capital, setting the development agenda. Regarding capital, the volume of concessionary lending declines relative to hard loans, while pov- erty increases.’ Net transfers from South to North* show that the current system is not working as well as it should.° Regarding ideas, there is profound confu- sion, which development agencies could help greatly to clarify. Development agencies are not primarily responsible for this situation, but they did not use as much of their considerable potential influence to change the conditions that contributed. Net transfers from South to North persist because of mature high-interest debt servicing, in spite of higher real interest rates in developing countries (averaging 17% during the 1980’s) (Human Development Report of the UN (HDR), 1992), compared with the 4% rates in OECD countries. To avoid negative transfers, more loans are needed just to cover debt service, thereby increasing total debt. The projects the debt supported were not as productive as expected; therefore, growth in debtor countries in the aggregate was less than expected. Conse- quently this repayment transfer is not made from a larger income made possible by the productivity of the projects financed by the debt. Of course, not all projects were disappointing, but on average, development prescriptions have not worked as well as calculated. This suggests that traditional prescriptions of how the North should help the South merit overhaul. The alternative view is that the North should stabilize 1ts resource consump- * One need is to calculate what fraction of the $55 000 000 000 total annual ODA supports sustainable investments and to increase this expectedly small fraction (disaggregating the large armaments fraction). Similarly for the possibly $100 000 000 000 per year from philanthropic grants. El Serafy’s sustainability method (1991) and environmental accounting (Ahmad, El Serafy, & Lutz 1989) should be widely used to unmask liquidation of natural capital assets. > The 1990 World Development Report “Poverty” calculated that more than one billion people, about one-third of the total population in the developing countries, live below the poverty line, and that poverty is also increasing in relative terms (World Bank, 1990). * For example, 1989’s South to North financial flows largely debt servicing and loan repayment approxi- mated $50 000 000 000, or $150 000 000 000 by lost trade, excluding brain drain costs. > The reasons why forty years of North-to-South capital transfers have not been as successful as planned are: a) improper allocation of capital, including government expenditures; b) flawed governmental policies that promoted misallocation, inefficient industries, and urban affluent elites at the expense of the rural sector; c) large and corrupt bureacracy and military; d) neglect of peasant agriculture; e) social systems that doom three-fourths of the population, especially women, to an unproductive and stagnating existence, especially failure to disseminate effective family planning. We are grateful to Professor Raymond Mikesell for this clarification. NORTHERN INCOME - SOUTHERN POVERTY 39 tion and reduce its damage to global life-support systems. Any higher consump- tion must come from productivity improvements, rather from increased throughput growth, a quantitative increase in size by the accretive of materials. If natural resources® were infinite, then growth would be unreservedly good. Since resources are finite, then more Northern growth inevitably means less room for Southern growth.’ Productivity improvements must replace through- put growth as the path of progress for the North, and eventually for the South as well. The purpose of this somewhat polemical paper is to contrast the two views, and to argue the case for more attention to the alternative view. The case merits a monograph, rather than a dozen pages. In being brief, we have not been able to deepen all the arguments. Our aim Is to raise the alternative view higher on the agenda so that monographs by others will be commissioned, and that the con- troversy over current doctrines and confusions is reconciled. Traditional View The North must grow faster to buy ever more resources from the South; otherwise the South will stagnate. Northern income growth translates into more Northern consumerism. Northern foreign exchange paying for imports from the South will indirectly trickle down from the Southern elites to alleviate poverty.® | UNDP’s 1992 Human Development Report outlines the historic discrediting of the trickle-down theory. The South is supposed to be almost totally dependent on the North and incapable of transforming its own resources into necessities for its own people. It must export natural resources, whose world market prices have, in general, steadily declined over the last few decades. The increased flow of natural resources supports Northern consumerism. These exports are for foreign exchange used partly to import the latest consumer goods for its own elites, who are not content with locally produced basic wage goods. If the econ- omy were unbounded by a finite ecosystem then this strategy would be possible and could be defended at least as the lesser evil. Although “‘trickle down” may not be thought the best means of achieving development, this view is widely held, and is espoused by development agencies and orthodox economists. ® Resources include the environment as a source of raw materials, healthy air etc., and as a sink for wastes, such as carbon dioxide. 7 For a discussion of environmental finitude and of sustainability in general, see Daly and Cobb (1989), Daly (1991), Goodland, Daly and El Serafy (1991), and Goodland and Daly (1992). The two views are contrasted best by Korten (1991), most pithily by Brooks (1991). For the most recent support for the alternative view, see Krabbe and Heijman (1992). Adams (1991) and George (1990) highlight the weaknesses of international development. ® For example, according to World Bank Vice President and Chief Economist Lawrence Summers (1991) ‘. . . rising tides do raise all boats.”” Rising Northern tides, however, imply ebbing Southern tides. 40 GOODLAND AND DALY Alternative View The North should stabilize its rate of consumption of resources to free them for the South, and to free up ecological space as well. The North has to reduce its overuse of global commons. Environmental sink capacity, and to a lesser extent environmental source capacity’ (Meadows et al., 1974), has been preempted by the North, thus denying as much room for the South. The North can continue to develop, but must cease increasing throughput growth. If the expanding global economy is bounded by a finite inexpandible ecosystem (Figure 1), then this view becomes the realistic one, and the traditional view becomes impossible. Foreign exchange generated by economic development, both from loans and exports, serves the desires of the rich more than the needs of the poor. Develop- ing countries should be more capable of producing necessities for their own people than of producing luxuries for their rich. The foreign exchange is needed more for the latter than the former. This minority alternative view is held by Economics Nobelists Jan Tinbergen and Trygve Haavelmo, along with many, if not most, of the members of the International Society for Ecological Economics. Tinbergen and Hueting (1991) hold that “. . . continuing (with the) prevailing growth path is blocking (global) chances for survival... .” Hueting (1990) says, ““. . . What the world needs /east is an increase in national income,” and ‘“*. . . the highest priority is to (halt) any further production growth in rich countries. . .” Haavelmo and Hansen (1991) characterize the two views: “Policies for more equality invariably start off with the statement that the standard of the poor should be raised towards the level of the rich. In other words, lifting the bottom rather than lowering the top.” The alternative view suggests adding “lowering or at least transforming the top’, ie: reducing Northern throughput growth and decreasing Northern consumerism. Under current dependency arrangements, a sharp Northern recession would hurt the South while benefiting the global envi- ronment. We advocate loosening such dependency to help prevent damage to the South.'° Discussion The traditional view and the alternative view cannot both be nght. The alter- native view leads us to emphasize the following overlapping elements, which ? Environmental limits to growth can be separated into source limits, such as depletion of petroleum, copper, etc., and sink limits, such as greenhouse effect, ozone shield damage, pollution, etc. '0 We could add a more palatable modification of the second view, an attack on today’s main environmental threat, namely pollution, by means of effluent charges, standards, etc. This would then be digested efficiently by the market and one indirect result would likely be a reduction in GDP and throughput. However, this falls into the obscurity of Brundtlandism: that the world needs a “5- to 10-fold increase in growth, but of a different kind.” While we would support such a frontal attack, we prefer to be crystal clear and opt for a transition away from throughput growth and towards a stable or declining throughput per unit of final product, and for a stable or declining population. NORTHERN INCOME - SOUTHERN POVERTY 41 Subsystem Population and Goods Produced } Finite global ecosystem Population and Goods Produced RECYCLED Fig. 1. The finite global ecosystem relative to the growing economic subsystem. The upper figure suggests the idyllic and long vanished era in which the economic subsystem was small relative to the then largely empty global ecosystem. The lower figure reflects today’s situation in which the economic subsystem is large relative to the global ecosystem. It is now basically full and is being stressed by the scale of the economic subsystem. 42 GOODLAND AND DALY together constitute our “‘ten reasons” why Northern growth is not the answer to Southern poverty. 1. GNP: A Flawed Measure of Human Well-Being GNP as conventionally measured can be a misleading guide in two ways. First, GNP has little to do with human welfare, as well demonstrated by UNDP’s 1992 Human Development Report. Second, economic sectors contrib- uting most to GNP are those that are the most environmentally damaging (see below). Although GNP-maximization is unreliable for both prudent economic development, as well as for prudent environmental management, economic development takes GNP-maximization seriously as a general goal or yardstick. This should not condemn economic activity properly directed to pollution abatement, conservation, and reducing waste. Recent work on environmental accounting by the Bank (Ahmad et al., 1989), Hueting (1990, 1992) and others shows that environmentally benign activities usually contribute a much smaller part to national income than do environmen- tally malign ones. On one hand, conventionally measured environmental dam- age and its rectification are “good” for GNP-boosting growth: for example, the Valdez oil spill clean-up boosted GNP. On the other hand, environmentally benign activities tend to be less costly than environmentally malign growth, consequently contribute less to GNP.'' For example, walking, biking and mass transit contribute less to GNP than does automobile use; train contributes less than airplane; an extra blanket or sweater less than raising the thermostat; one- child families less than six-child families; eating legumes less than eating beef; recycling less than trashing. Reduction of GNP resulting from choice of these benign activities should be encouraged, not resisted. Therefore, environmental protection is not, as commonly portrayed, an ex- pensive choice, largely to be chosen when a nation becomes rich enough to afford such choices. The opposite is true. At the same time, rectifying environ- mentally harmful growth is indeed staggeringly expensive: for example, nuclear and toxic cleanups or greenhouse effect reversal. This strengthens the argument for prevention rather than cure, and for not repeating the errors of the industrial countries which passed through an environmentally damaging phase of eco- nomic development. The contrast between Northern and Southern environments is that much local Northern environmental damage is pollution, hence reversible. For exam- ple, London’s River Thames pollution and the “‘pea-soup”’ smogs of the 1950s '! For evidence and arguments supporting this important conclusion, see Hueting et al. (1992), appendix 3. NORTHERN INCOME - SOUTHERN POVERTY 43 have been largely reversed. On the other hand, most Southern environmental damage is irreversible loss of biodiversity.'* Irreversible damage cannot be cured; replacement costs are infinite. The North’s unnecessarily expensive “damage the environment, then cure it” approach may be affordable (but im- prudent) for the rich North. But cures cannot work for irreversible damage, and the South could not afford such expensive approaches in any event. The preven- tive approach is the only possible one for the South. 2. Importance of Relative Incomes The traditional view emphasizing global income growth will exacerbate in- equality while scarcely denting poverty. An annual 3% increase in global per capita income translates initially into annual per capita increments of $633 for the US, but only $10 or less for China, India, Bangladesh, or Nigeria. After a ~ decade, the US income will have risen by $7257, whereas such income growth will have raised Ethiopia’s income by only $41. Therefore, advocates of the traditional view, prioritizing global income growth, should at least state that an unwanted side effect will be to worsen income disparity. When dealing with market competition for finite resources, relative income is more influential than absolute income in determining whether some individuals are excluded from access to available resources. Since markets need at least some social equity, the traditional view will gradually exclude the poor from domestic and international market economies. We emphasize equity within countries as well as between countries. 3. Differential Utility of Needs and Wants For Northern consumers, self-evaluated happiness is more a function of rela- tive income than of absolute income. Therefore, since aggregate growth in- creases absolute but not relative income, it contributes little to actual happiness in the North (Hirsch,- 1976). So, although our main concern is alleviation of absolute poverty, we recognize that above the poverty level, relative income is a more important determinant of satisfaction than absolute income. Now that Northern income growth yields sharply declining marginal utility, the North should question whether raising its incomes will not increase environmental costs faster than it increases production benefits. Raising Northern incomes not '? This generalization stems from the orders-of-magnitude-richer biodiversity in tropical countries and by the related lack of tropical winters. The four main tropical environmental impacts—deracination of jungle dwellers, deforestation, extinctions, and topsoil loss—are irreversible. Water and air pollution in the North are basically reversible. Pervasive global negative externalities (e. g., carbon dioxide accumulation) are probably irreversible over most time frames. The operational distinction between reversible and irreversible damage is that cure for the former is possible for the rich; prevention is the only choice for irreversibles. 44 GOODLAND AND DALY only widens the gap between North and South, but may well be reducing North- ern welfare absolutely. In the North’s choice between consumerism and saving, the quest for relative standing based on visible commodities has biased the North towards consumerism. With less consumerism, more saving in the North could be invested in much needed poverty alleviation and growth in the South. Production to meet basic human needs produces relatively high utilities, fre- quently with relatively low environmental costs. Wants or luxuries generate relatively lower utility, often with with higher environmental costs. 4. Misplaced Technological Optimism New technology is often adopted in order to improve productivity, which in turn can raise material standards of living. The impact of a particular technology depends on the nature of the technology, the size of the population deploying it, and the population’s level of affluence. In the I = PAT identity, impact equals population times affluence times technology.'* Accept here as given that world population is projected to double in 40 years and that the rich countries’ per capita income ($18,330) is 23 times that of the poor and middle income coun- tries ($800).'* Therefore, to raise Southern affluence to today’s level of the North (holding both impact and Northern incomes constant) means technology must improve 2 X 23 or 46 times. Since historical technological improvement rates never have exceeded a fraction of the needed 46 times, it will be exceedingly difficult for poor countries to catch up with rich countries in 40 years even if the North maintains current levels of income. It will be that much more difficult if the South is to catch up with a moving target, as prescribed by the tradi- tional view. Furthermore, this 46-fold increase must be in resource efficiency, and not just in capital or labor efficiency. Historically, much of the increase in capital and labor efficiency has been at the expense of resource efficiency. In agriculture, for example, the increase in labor and capital productivity has required an enor- mous increase in the complementary resource throughputs (energy, fertilizer, biocides, water) whose productivity has fallen. 5. The Value of Economic Self-Reliance The poor can be helped far more, and with much less environmental damage, by a pattern of development which promotes employment in developing coun- '3 Impact here means impact on or damage to environmental sources or sinks; affluence means per capita consumption of resources; technology refers to technological efficiency defined in terms of the number of units of human well-being produced per unit of environmental cost. Thus, where I is impact, P is population, and Y is total production, then I = P x Y/P X I/Y. '* Data from The World Bank’s World Development Report 1991, Table A.2. | | | NORTHERN INCOME - SOUTHERN POVERTY 45 tries—as recently advocated by the World Bank’s 1990 “Poverty” WDR, rather than by increasing Northern consumption and relying on “trickle down,” as advocated by the traditional view. Poverty alleviation needs employment and self-reliance strategies aimed at using local resources to produce for domestic needs. This translates partly into promotion of value-added and domestic pro- cessing, and partly into employment creation. True, developing countries may at first waste a large fraction of raw materials during processing because of using obsolescent technology commonly transferred to the South. For example, mod- ern sawmills waste considerably less wood than obsolescent sawmills do. But this argues for accelerating transfer of up-to-date technologies, rather than the old colonialist approach of exporting raw materials to be more efficiently milled in the North. The most needed such technologies are renewable energy genera- tion and contraceptive methods. Waste prevention, recycling methods, pollu- ‘tion prevention, efficiency increases (e. g., of sawmills), low-input, organic and recycling agriculture, and methods reducing material- and energy-intensity in manufacturing also are priorities. Duchin (1992) argues for supplementing tech- nology transfer by the practice of industrial ecology: life-cycle engineering for reducing pollution. 6. Throughput Growth as a Source of Both Income Growth and of Environmental Damage If the activities contributing to national income are disaggregated into two components, environmentally friendly (e. g., most government services), and environmentally burdening (e.g., industry, agriculture, utilities), then about one quarter of the activities (measured in labor volume) generate about 65% of increases in national income. “Unfortunately, that 25% is precisely the activities which impair the environment.” (Hueting et al., 1992) Increases in productivity generated by a relatively small part of the economy spread over the whole society via labor supply demand linkages. For example, a barber’s labor volume and real output have not appreciably increased over the last 40 or 100 years, but his (deflated) income or value added has risen by a factor of four. The barber’s increased real income has been generated by activities other than his own. These other activities are much harder on the environment than his own activities. Average Northerners now consume vastly more than they did 40 years ago all the way up the income scale—more than twice as much in the case of the US and Japan. For example, 88% of US households now own at least one car (up from 55% in 1935), and the average number of vehicles per household is two—even for barbers. 7. Subsidized Resource Pricing The poor can be supported more directly and with less environmental damage by “getting the price right,” or at least getting the price better than at present. 46 GOODLAND AND DALY Today’s severe undervaluation of Southern raw material exports means the South is subsidizing the North, in the sense of both externalized environmental costs, as well as governmental incentives, such as logging road construction. Cheap tropical log exports are a case in point. Stupendous subsidies, in the form of unpaid environmental costs, are only beginning to be recognized. Eastern Europe’s pollution is a case in point. In the absence of Southern cartels or of producers’ agreements to limit production, unilateral price changes are unlikely. We therefore advocate that international organizations, such as the World Bank, the IMF, or the UN, should foster and promote more economically realistic full-cost pricing. A caveat: the North advocates removal of subsidies, but this may hurt the poor more than the rich because in developing countries, the rate of removal of subsidies to the poor exceeds the rate of removal of subsidies to the rich. Full-cost pricing should also be used to encourage the South to exploit its comparative advantage in agriculture, labor-intensive industry, and raw mate- rial processing in order to increase its employment, modernize its subsistence sector, and raise its per capita incomes. 8. Inequitable Trading Systems “The structure of trade . . . is a curse from the perspective of sustainable development.” (Haavelmo and Hansen, 1991) The writers conclude: . . . Much Northern growth is based on depleting Southern resources for a price far below the cost of sustainable exploitation. The adoption by the North of the “full cost” principle for pricing Southern resources would help the South more than would Northern growth. Exports only serve a purpose if they finance useful imports. The North should not (tell) the South to export what it cannot afford. Strategies to enhance exports of many staple agricultural prod- ucts should be critically revisited. Such goods face low demand elasticities in world markets. Individually each exporter takes the world market price as given. In the aggregate, however, the simultaneous implementation of such strategies by many drives the price down dramatically as they all reach their production targets. In the end the export revenue might fall short of paying for the imported machinery, implements, pesticides etc., required (to produce) for export. 9. Dysfunctions of Imbalanced Trade The traditional view tends to overestimate the virtues of free trade—that is, deregulated commerce across national boundaries. Financial imbalances from deregulated trade have led to debts that are unrepayable, and attempts to repay them by rapid export of raw materials can be environmentally destructive. Natural resource stocks are liquidated to meet debt servicing flows. Current efforts under GATT to include services under free trade will subject that sector NORTHERN INCOME - SOUTHERN POVERTY 47 to international competition further pressuring existing payment imbalances. There is a conflict between the “free trade prescription” and the “get the price right” prescription. Countries following World Bank advice to internalize exter- nal environmental costs should not be expected to engage in free trade with countries that do not follow similar rules of cost internalization. Tariffs to pro- tect an efficient national policy of cost internalization (not an inefficient in- dustry) should not be ruled out as unwarranted “protectionism.” Unpaid envi- ronmental costs, such as liquidation of natural capital, are subsidies reducing the price of exports—tantamount to dumping. User costs, from this point of view, should internalize depletion of natural capital. Rectification of the asym- metry of anti-dumping laws for manufacturers, but not for raw materials, would promote global sustainability. This refers to US Pacific Northwest logs exported to Japan as well as Malaysian rainforest hardwoods exported to Europe. At the same time, we acknowledge that Southern trade policies have limited intra- South trade in goods and services, which need to be expanded, and have contrib- uted to real transfers from the rural to urban sectors. 10. The Insecurity of Inequality From the ecological-economic point of view, our main concerns are that the prescription of raising Northern income will fail to alleviate poverty, will worsen inequality, and will reverse current trends towards sustainability. To these eco- logic-economic aspects we append a final concern, that of global security. We believe raising northern incomes will decrease global security, and in HE (His Excellency) Minister Salim’s Indonesian view tend to foment social stress and even revolution. Specifically, decreasing sustainability will increase “‘environ- mental refugees’—those people forced out of their homes and countries by environmental mismanagement, man-made disasters, and development-in- duced expulsions, including poisoned water or air, eroded soils, and desertifica- tion. A specific example is the environmental damage from Papua New ’ Guinea’s Panguna copper mine, which was a major cause of the recent civil war in that country. The North bears an overwhelming responsibility for many environmental costs to both sinks and sources. As stated in the 1991 Beijing Declaration, the South has taken note that the North is responsible for practically all historical global pollution and continues to emit most of today’s global pollution. Some Southern writers (Agarwal and Narain, 1991) argue that the North owes repara- tion payments to the South for historically disproportionate pre-emption of the global commons. Reparations are to restore base-line equilibria. Northern secu- rity will be enhanced to the extent the North reduces inequality in the South. North-South environmental linkages are growing. Two examples: While East- 48 GOODLAND AND DALY ern Europe pollutes Scandinavian air, Scandinavia finds it more cost effective to improve its air quality by financing pollution abatement in Eastern Europe itself rather than in Scandinavia. Similarly, the Netherlands finds it more efficient to sequester Amsterdam-produced carbon dioxide from its coal-fired thermal elec- tricity plants by financing tree plantations in South America. South America gets much employment created and more wood construction materials; the Netherlands buys time to internalize its own wastes within its own borders. The North should see it as in its own direct security interest to invest in the South to reduce inequality, to alleviate Southern poverty, to protect and im- prove the global environment, and to avoid creating environmental refugees. The South can raise funds by taking on board some of the North’s environmen- tal concerns (such as carbon-sequestering in which the tropics have a major additional advantage) by means of tradeable quotas, and by selling the benefits gained from use of their environmental assets, such as intact tropical forest. The traditional view wants to appropriate more Southern resources for the North. The alternate view 1s that the North has to learn to live within its own means, to reduce its current reliance on global commons and on the environmental re- sources of the South. Conclusion A corollary of the above argument is that possibly the best way in which the North could help the South is by adopting the first oath of Hippocrates: “‘First, cease doing harm.” The traditional view exacerbates harm; the alternative view is more likely to help. In the global approach to sustainability, the North has to adapt far more than the South. The South’s contribution to global sustainability could be population stability and prevention of irreversible losses. The North’s contribution to environmental damage in the South is clear: ozone shield dam- age, climate change, greenhouse warming, tropical deforestation, export planta- tions forcing the poor to marginal lands, indebtedness promoting drawdown of natural capital, and overuse of potentially renewable resources. Recommendations The ways in which the North can help the South, in priority order, are as follows: | First, the North must get its own house in order by transforming today’s Northern consumerism and borrowing economy into a more sustainable model. An accelerated transition to renewable energy for a stable population is the major element. Internalizing environmental costs in energy prices would be a powerful start. This transition has to be faster than what would be suggested by the market at present. OECD countries are unilaterally phasing in carbon-based NORTHERN INCOME - SOUTHERN POVERTY 49 or non-renewable energy taxes. The implications of such taxes for Southern development should be discussed. Part of Northern energy taxes could be allo- cated to promote Southern sustainable development. Such national taxes may later become global, as proposed by UNCED in 1991, or become tradeable pollution permits (perhaps with futures and options), as proposed by UNCTAD in 1992. Such arrangements would help protect global life-support systems. Second, the North should internalize the costs of disposal of its toxic and other wastes within its own national borders, rather than exporting them in the name of “comparative advantage” to low wage countries. Internalization of costs to the nation of origin, as well as the firm, gives a stronger incentive to minimize toxic waste generation. This dynamic benefit is more important than the static allocation cost of neglecting ““comparative advantage.” Third, the North should halt the harm to the South inflicted by present poli- -cies. This includes underpricing of Southern exports, warfare, and global pollu- tion. Of course, the South has a bigger role to play in solving its own problems (as outlined in footnote 5). Fourth, Northern governments, the private sector, and development agencies create much Southern debt, much of which is unrepayable. The North should address the current imbalance between commercial rate loans, subsidized in- vestments, and grants to the South. The relative proportions of Northern transfers—as (1) loans, (2) subsidized, almost concessionary, IDA-type arrange- ments, or (3) grants such as that of free access by the South to the North’s socially and ecologically beneficial technologies—should be improved. What are the proportions now? More grants and subsidized loans relative to hard loans. Re- parations are mentioned above. Questionable loans—loans accelerating liqui- dation of natural capital, loans failing to internalize full costs, unrepayable loans, and loans clearly for unsustainable purposes—should be halted or can- celled. Global sustainability, poverty, equity, and security would be improved if debts in severely indebted countries were conditionally written off commensu- rate with progress towards environmental sustainability. Fifth, because economic justification for foreign exchange loans for environ- mental investments is difficult, the conventional cost-benefit analysis needs to be broadened to internalize more environmental costs. Where international assistance is required for the South’s global or transnational environmental priorities, it should be grant funded. Recent World Bank improvements in this respect are encouraging and need to be accelerated. Economists should begin to consider environmental investments as extended infrastructure investments— in other words, investments in the maintenance of the biophysical infrastructure that supports all economic activity, both public and private. Therefore, where conventional cost-benefit analysis is difficult to apply, as in some conventional 50 GOODLAND AND DALY infrastructure investments, the Bank, UNDP and UNEP can now make envi- ronmental grants through the pilot Global Environmental Facility. This impor- tant facility urgently needs to be revised and vastly expanded if it is to help the South to approach sustainability. Sixth, the North should focus on direct help to the South rather than on indirect “trickle down” help. Investments should focus more directly on the poor countries, and on the poor strata of society in those countries. Less ODA (Official Development Assistance) should be on commercial (tied) terms, and should include only the most essential projects, emphasizing domestic needs more than the export market. Suggestions to finance such investments have included reparations, conditioned debt relief (e. g., Brady Plan, Trinidad Terms), subsidized loans, and especially grants. The investments are to acceler- ate needed growth and employment-creation in small Southern economies. International assistance is needed to facilitate and possibly to help purchase rights to environmentally beneficial technology for the South. This will include improving the policy framework for commercial transfer of technology, and training and institution strengthening to improve absorptive capacities. As old Northern assets depreciate, some may better be replaced in the South with appropriate technology. Thus, the North should accelerate export of ad- vanced but appropriate technology for Southern processing of their raw mate- rials. If the North finds cleaner technology easier to espouse than reducing population or overconsumption, then the North ought to be more willing to transfer technology on easier terms, even at the expense of other ODA. The traditional view would argue for more Northern growth in order to increase ODA. GATT needs to start addressing the environmental implications of trade and needs to take on board the concept of sustainable development. Seventh, the priorities for sustainable economic development in the South are: a) Accelerate the transition to population stability. b) Accelerate the transition to renewable energy. c) Human capital formation: education and training, employment creation, particularly for women. d) Technological transfer to leapfrog the North’s environmentally damaging stage of economic evolution; job creation rather than automation. e) Direct poverty alleviation, including social safety nets and targeted aid. Former World Bank President R. S. McNamara concluded his 1991 United Nations address by calling for official discussion of “. . . how the developed world, consuming seven times as much per capita as do the citizens of the developing countries, may both adjust . . . consumption patterns and reduce the environmental impact of each unit of consumption, so as to help assure a NORTHERN INCOME - SOUTHERN POVERTY 51 sustainable path of development for all the inhabitants of our planet. It will be neither morally defensible nor politically acceptable to do less.” Acknowledgments We acknowledge with pleasure the help of our World Bank colleagues, espe- cially Salah El Serafy, as well as that of Johan Holmberg, Roefie Hueting, David Korten and Raymond Mikesell. References Adams, P. (1991). Odious debts. London: Earthscan. Agarwal, A., & Narain, S. (1991). Global warming in an unequal world. /nternational Journal of Sustainable Development, 1:98-104. Ahmad, Y., El Serafy, S., & Lutz, E. (Eds.) (1989). Environmental accounting. Washington: The World Bank. Brooks, D. (1991). An evaluation of “Our Common Future.” Human Economy Newsletter, 12:4. Daly, H. E. (1991). Ecological economics and sustainable development: From concept to policy. (Environment Department Paper 1991-24.) Washington: The World Bank. Daly, H. E., & Cobb, J. (1989). For the common good: Redirecting the economy towards community, the environment, and a sustainable future. Boston: Beacon Press. Duchin, F. (in press). Prospects for environmentally sound economic development in the North, in the South, and in North-South economic relations: The role for action-oriented analysis. Journal of Clean Technology and Environmental Sciences. EI Serafy, S. (1991). The environment as capital. In Costanza, R. (Ed.), Ecological economics (pp. 168-175). New York: Cambridge University Press. George, S. (1990). A fate worse than debt. New York: Weidenfeld. Goodland, R., & Daly, H. E. (1992). Approaching global environmental sustainability. Society of International Development, 2:35-41; 3:64-71. Goodland, R., Daly, H.E., & El Serafy, S. (1991). Environmentally sustainable economic development: Buila- ing on Brundtland (Environment Paper 36). Washington: The World Bank. Haavelmo, T., & Hansen, S. (1991). On the strategy of trying to reduce economic inequality by expanding the scale of human activity. (27-35) In Goodland, R., Daly, H., and El Serafy, S. (Eds.), Environmentally sustainable economic development: Building on Brundtland (Environment Paper 36). Washington: The World Bank. Hirsch, F. (1976). The social limits to growth. Cambridge: Harvard University Press. Hueting, R. (1990). The Brundtland report: A matter of conflicting goals. Ecological Economics, 2:109-117. Hueting, R., Bosch, P., & de Boer, B. (1992). Methodology for the calculation of sustainable national income (Statistical Essays, M series). Den Haag: Central Bureau of Statistics. Korten, D. C. (1991). Sustainable development. World Policy Journal, (Winter 1991-1992):156-190. Krabbe, J. J., & Heijman, W. J. M. (Eds.) (1992). National income and nature: Externalities, growth and steady state. Dordrecht: Kluwer Academic. McNamara, R. S. (1991). Global population policy to advance human development in the 21st century. New York: United Nations. Meadows, D. H., & Club of Rome. (1974). The limits to growth. New York: Universe. Ministerial Conference of Developing Countries on Environment and Development. (1991). Beijing ministerial declaration on environment and development. Beijing: Author. Summers, L. (1991). Research challenges for development economists. Finance and Development, 28:2-S. Tinbergen, J., & Hueting, R. (1991). GNP and market prices: Wrong signals for sustainable economic success that mask environmental destruction. In Goodland, R., Daly, H., & El Serafy, S. (Eds.), Environmentally sustainable economic development: Building on Brundtland (pp. 36-42) (Environment Paper 36). Washing- ton: The World Bank. UNDP. (in press). Human development report. New York: UNDP. World Bank. (1990). World development report: ““Poverty.”’ Washington: World Bank. Instructions to Contributors Revised March 1993 Supersedes all earlier versions Objectives of the Journal The purpose of the Journal of the Washington Academy of Sciences is to publish the results of original research, critical reviews, and historical articles that contribute significantly to knowledge in all areas of science and engineering. Articles should be intelligible to readers in a variety of disciplines. / Submission of Manuscripts Manuscripts should be addressed to Editors, Journal of the Washington Academy of Sciences, Department of Mathematics & Science, Mount Vernon College, 2100 Foxhall Road NW, Washington DC 20007-1199, USA. Contributions appearing in the Journal of the Washington Academy of Sciences are normally original papers which have not been published elsewhere. The publication of a paper that has been widely disseminated is permitted only if the editors judge that the manuscript contains amplification or clarification of the original material of significant benefit to the scientific community. Prior appearance of material reported in the submitted manuscript must be noted in the cover letter. It is the obligation of the author(s) to inform the editors if there are any circumstances about the contribution that bear on this policy. The style of manuscripts should follow the model of the Publication Manual of the American Psychological Association, 3rd edition, with a few exceptions. The main exception is that text which is intended to be italicized should use word processor italic code rather than underline code. A second exception is in the use of numbers (see below). Manuscripts may be submitted in either the traditional paper format or electronically. WordPerfect 5.1 is the preferred word processor. In either case, the submission will be acknowledged by the editors. With paper manuscripts, an original and three clear review copies of everything must be submitted. Each copy should include glossy or clear photocopies of all figures. Photocopies are acceptable only if they show all details necessary for critical examination of the data. This is particularly important for halftone photographs. The editors encourage contributors to use a grammar- and style-checking program such as RightWriter, Grammatik, or Editor before submitting their manuscripts. Poorly written manuscripts will be returned to the authors for rewriting. General Policies and Guidelines Review Procedures The editors and associate editors are appointed by the Washington Academy of Sciences and have final responsibility for all editorial decisions. Reviewers are selected by the editors, in consultation with associate editors, for their competence in a specialized area of science or engineering. When a manuscript is received, the editors will first judge if its content falls within the scope of the Journal of the Washington Academy of Sciences. Manuscripts of a very 32 specialized nature, or those that in the judgment of the editors are not of sufficient interest to the general reader of the journal, will be returned to the author(s) without review. If the manuscript is acceptable, it will be sent to two independent reviewers for evaluation. The reviewers are advisory to the editors and their reports are used to reach all editorial decisions. If the reviewers disagree, or if in the judgment of the editors the manuscript has not received adequate consideration, the manuscript will be given to a third reviewer. When a manuscript is returned to the author(s) for revision, they should reply to all specific recommendations of the reviewers in an accompanying letter, indicating the recommendations they have incorporated into the revision and their reasons for disregarding those they feel are unacceptable. Minor corrections may be made on the manuscript by typewriter; major corrections will require retyping of the entire page or section. Handwritten corrections will not be accepted. The revised manuscript, in triplicate, should be returned to the editors within 60 days. Revised manuscripts received more than 120 days after return will be considered as new submissions and will undergo a new review. English Language To assure the widest possible readership, only manuscripts in English will be considered for publication. Authors not entirely familiar with English usage should seek a native English- speaking colleague for advice on correct syntax and word usage. English-language editorial assistance is available when needed. Correct style and word usage, however, are the responsibility of the author(s). References in the Text All reference material should be accessible to the public (i.e., articles in standard journals and governmental reports). References should follow APA style: they are cited in the text by author and year (Fenchal & Finlay, 1992), or by year alone when the author’s name is in the text; for example, "as described by Olive and Blanton (1980), ... .". Unpublished experiments or observations, personal communications, papers submitted for publication or in preparation, abstracts and dissertations should not be included in the References section but should appear parenthetically in the text. Hyphenation Do not use a hyphen to divide words, including compound words, at the end of a line; if the words are unfamiliar to the printer, they may be incorrectly hyphenated. Compound words consisting of a noun and an adjective should be hyphenated, as in electron-microscopic examination. Noun-noun compounds used as adjectives should also be hyphenated, as in lymph- node cells. Right-Justification Do not justify right margin of text whether or not your word processor permits you to do this. Boldface and Italics Word processor code is sufficient. If your word processor does not have boldface and italic codes, then underline text to be italicized and put a wavy underline under text to be boldfaced. 53 Metric System The metric system should be used in all instances. Use SI units with standard prefixes. (In 1990, the prefixes yocto- (y) (denoting 10%‘), zepto- (z) (denoting 107'), zetta- (Z) (denoting 10?'), and yotta- (Y) (denoting 10”) were added to the SI system.) Do not use nonstandard units such as Angstrom, millimicron, cubic centimeter, etc.; rather use nanometer (nm), microliter (uL), etc. Compound units, such as newton-meter, should be denoted N m or N-m (using WordPerfect character 6,32, the small center dot). Quotients of units, such as meter per second, should be denotea m/s, m s", or m:s". Preparation of Manuscripts General Instructions The instructions for the development and submission of a manuscript should be followed exactly to prevent delays in reviewing and handling. Before preparing a new paper for submission the author(s) are advised to examine carefully current issues of the Jourxal to familiarize themselves with the Journal’s conventions and note any changes in style. All pages of the manuscript must be typewritten on one side of the paper, double-spaced throughout (including Acknowledgments, References, Footnotes, Tables, and Figure Legends) on 22 x 28 cm (8'4" x 11") bond or heavy-bodied paper. Ample margins of about 2.5 cm (1") should be left on all sides for editorial corrections and queries. Number the manuscript pages consecutively in the upper right corner beginning with the title page. The author’s name should not appear on the manuscript beyond the title page. The manuscript should be prepared with a standard typeface, size Elite or larger, using a new black ribbon and solid white paper. Computer generated: documents must be clear and readable of letter quality, using a open dot matrix or laser printer. The author is responsible for putting the paper into the correct form. If this is not done at the outset, changes in the manuscript or text will have to be made during refereeing and editing, resulting in delay of publication. The author(s) should use clear, concise English. It is often useful to have a draft of the paper read by one or more colleagues. They will often be able to point out irrelevant matter and illogical or verbose statements. While a polished literary style is not demanded of scientific papers, they should conform to the elementary rules of grammar, syntax, and punctuation. Unattached participles, subject and verb complements, and long sequences of nouns are some of the most common errors that author(s) should look for in their writing. Words which may not be familiar to most readers should be boldfaced and defined the first time they are used. Slang and language that reflects prejudice (e.g., sexist language) must be avoided. Cover Letter A submittal letter must accompany the manuscript and should include a statement by the author(s) that no significant amount of material reported has been published, or is under consideration for publication elsewhere. Permission to Reprint The author must obtain written permission to reprint illustrations or tables from other publications. Any such material should be accompanied by a complete citation of the copyright owner (usually the publisher) and a statement that permission to reprint has been granted. 54 ——— eee Organization The text of the manuscript for an original experimental contribution should be arranged as follows (not all sections need be included): (1) Title Page; (2) Abstract; (3) Introduction; (4) Materials and Methods; (5) Results; (6) Discussion; (7) Acknowledgments; (8) References; (9) Footnotes; (10) Table Legends; (11) Tables; (12) Figure Legends, followed by art work and camera-copy tables. Critical reviews and historical articles should adhere to the general format of regular published papers in the Journal but do not require the subheadings of an experimental paper. The desired position of tables and figures should be indicated in the left margin of the text. Subheads All subheads should be descriptive clauses, not complete sentences or questions, and should follow the format shown below. One level: Level 1 Subhead Two levels: Level 1 Subhead Level 2 Subhead Three levels: Level 1 Subhead Level 2 Subhead Level 3 Subhead. Text. Four levels: Level 1 Subhead Level 2 Subhead Level 3 Subhead Level 4 Subhead. Text. Five levels: LEVEL 1 SUBHEAD Level 2 Subhead Level 3 Subhead Level 4 Subhead Level 5 Subhead. Text. Title Page Page one should contain a short "running title" (no more than 70 letters and spaces) at the top of the page; the title (no more than 100 letters and spaces); the author(s)’ name(s) (capitalize all letters and omit degrees or titles), and full departmental and institutional affiliation(s) (in italics); a list of key words not present in the title, and the name, address, telephone number and facsimile (FAX) number, if available, of the corresponding author. The title should be specific and informative. Subtitles are not encouraged and a pair of articles numbered "I" and "II" is acceptable only if both are submitted and accepted together. 55 Avoid special symbols and formulas in titles unless they are essential to indicate content. In the title, capitalize the first letter of all words except articles, prepositions, and coordinate conjunctions. The listing of the names of coauthors implies that the manuscript meets with their approval. The affiliation appearing under the author’s name should be the organization where the experimental work was carried out. If this is no longer the author’s current affiliation, the latter address is given in a footnote. Funding sources, contribution numbers, and presentations, if any, of the material given at a scientific meeting (give dates and sponsoring societies) should be acknowledged in a footnote. Abstract The abstract starts on page 2 and is a brief summary or synopsis consisting of 250 words or less. It should include information on objectives, procedures, results, and conclusions. When published, the abstract will precede the introductory section of the text. The abstract should be written in complete sentences; it should be self-explanatory and intelligible without reference to the rest of the article. The abstract should not be written in the first person and abbreviations and references should not be used. Introduction The introduction (not headed) should begin on page 3. The introduction should state the purpose of the investigation, briefly and clearly describing its relationship to previous research in the field and on related topics. However, extensive reviews of the literature should be avoided. Materials and Methods (Experimental Procedure) In order to enable full understanding and allow for the results to be reproduced and confirmed, the materials and experimental procedures section should describe in sufficient detail the experiments performed. Normal experimental procedures should be described in detail, with shorter terse descriptions given for general techniques and complete published procedures referred to by literature citations of both the original and any published modifications. When experimental procedure requires the use of particular products and equipment, the name and location of the supplier should be given in parentheses. Whenever hazardous procedures or materials are utilized, the necessary precautions should be stated. Results The results should be presented concisely. Tables and figures should be used only if they are essential for the comprehension of the data. The same data should not be presented in more than one figure, in both a figure and a table, or redescribed completely in the text. As a rule, interpretation of the results should be reserved for the discussion. Discussion The purpose of the discussion is to interpret the results and to relate them to existing knowledge in the field. In general, observe brevity consistent with clarity. Information given elsewhere in the manuscript should not be repeated in the discussion. Extensive reviews of the literature should be avoided. 56 + eS ee Acknowledgments Acknowledge in this section technical assistance, advice from colleagues, gifts, etc. Financial support should be acknowledged in a footnote to the title. ' References Articles "in press" may be cited in the reference list, but articles “submitted for publication" or "in preparation" should not be included. Journal Articles. Each citation to a journal article should give the surnames and initials of all authors, followed by the year in parentheses, title of the article, name of the journal, volume number, and inclusive pages. In titles of articles, capitalize only the initial letter and words that are capitalized in normal text. /talicize titles of journals, capitalizing major words. (Clear abbreviations are acceptable.) /talicize volume numbers. Examples: Brill, D., Ayres, V., & Gaunaurd, G. (1987). The influence of natural resonances on scattering and radiation processes. J. Wash. Acad. Sci., 77, 55-65. Kumar, S. (1987). Fatty acid synthase: a protein having many catalytically active domains. J. Wash. Acad. Sci., 77, 66-75. Articles in Books. Citations to articles in books should give the surnames and initials of all authors, followed by the year in parentheses, title of the article or chapters, editor(s)’ name(s), book title, inclusive pages in parentheses, place of publication, and publisher. In titles of articles and books, capitalize only the initial letter and words that are capitalized in normal text. J/talicize titles of books. (Do not abbreviate.) Example: Nozawa, Y., & Thompson, G. Y. (1981). Lipids and membrane organization in Tetrahymena. In M. Levandowsky & S. H. Hutner (Eds.), Biochemistry and physiology of protozoa (2nd ed., pp. 275-338). New York: Academic Press. Books. Citations to books should give the surnames and initials of all authors, followed by the year of publication in parentheses, title of the book, place of publication, and publisher. In titles of books, capitalize only the initial letter, any word following a colon and proper nouns. Italicize titles of books. (Do-not abbreviate.) Example: Riedl, R. (1978). Order in living organisms: A system analysis of evolution. New York: John Wiley. Technical Reports. Citations to technical reports should give the surnames and initials of all authors, followed by the year of publication in parentheses, title of the report, report number, place of publication, and publisher. In titles of technical reports, capitalize only the initial letter, any word following a semicolon or a dash, and proper nouns. /Jtalicize titles of technical reports. (Do not abbreviate.) Example: Brush, L. (1979). Why women avoid the study of mathematics: A longitudinal study. (Contract No. 400-77-0099) Washington, DC: National Institute of Education. Si Tables All tables must be cited in the text of the manuscript. Since tabular material is expensive to reproduce, it should be simple and uncomplicated, with as few vertical and horizontal rules as possible. Tables should be on separate pages assigned Arabic numbers and arranged consecu- tively. Table titles should be complete but brief. Information other than that defining the data should be formatted in paragraph style. Footnotes within tables should be indicated by super- script letters: *, °, °, etc. Indicate in the margin where the tables are to appear in the text. Figures (Illustrations) Figures should be assigned Arabic numbers and arranged consecutively. Illustrations that are to be grouped together should be mounted as plates in the most economically arranged layout. Individual figures within a group must be squared accurately and mounted with edges touching (butted) and with no white line showing between the figures. When several figures are grouped together, the figure numbers should be placed in the lower right corner whenever possible. All illustrations, whether layouts or single figures, may match either the size of a single column (63 mm) or the width of the entire page (126 mm). The maximum length of a layout is 201 mm. It is advisable to protect all illustrations and figures with non-abrasive dust covers. On the back of each illustration, indicate the figure number, author’s name, a brief title, and an arrow pointing to the top of the illustration. Halftone Illustrations. Halftone illustrations (that is, photographs, micrographs, shaded and wash drawings) should be sharp, well-contrasted glossy prints of the original negative, maximally trimmed at right angles to eliminate unnecessary areas without significant information content, and should be of the desired final size. It is important that author(s) use the maximum possible size only to increase the information content of illustrations; large peripheral areas of figures with low information content must be avoided. Original halftone illustrations must be mounted on heavy-weight drawing paper or thin poster board. Line Drawings. Line drawings should be submitted as black India ink drawings on white paper. High quality photographic reproductions (glossy prints) are acceptable. Line drawings and copies of figures need not be mounted. Figures (Legends) Legends should be concise and provide a brief, self-sufficient explanation of the illustrations. Legends for plates with multiple figures should be grouped together. Provide a title for the entire plate followed by the descriptions of the individual figures in the plate. If the magnification is quoted in a sentence, it should appear in parentheses; for example, (X 500). At the end of the legend, it should appear as a sentence fragment; for example, " cytoskeleton. X 41,000." Use the cross (WordPerfect character 6,39), not the letter x, to denote multiplication. The length of calibration bars should be given; for example, “Bar: 1.0um.” Please italicize “arrows,” “arrowheads,” “asterisk” when these words appear in a legend. 99 6¢ Footnotes All footnotes should be placed on a separate page. The first footnote should be from the title page and refer to the title of the manuscript and/or the author(s). Textual footnotes should include a list of abbreviations, be numbered consecutively, be brief, and be kept to a minimum. 58 Common Questions and Errors 1. Spell out acronyms and abbreviations the first time they are used unless they are common throughout the discipline. 2. In the text, avoid beginning a sentence with an abbreviation, a symbol, or a numeral. 3. A generic name should be spelled out the first time it is used but abbreviated any additional times it is used. Example: Euplotes eurystomus the first time; E. eurystomus the second time. 4. When reporting statistics, the name of the statistic, the degrees of freedom, the value obtained, and the p-value should be reported. 5. Use roman type, not italic, for chemical symbols for elements and chemical formulae. The mass number precedes the symbol. Example: '7C. 6. The standard symbols (without periods) for SI units should be used. If English units - (such as inches or pounds per square inch) are used, their metric equivalents must be given in parentheses. Abbreviations, especially of units, should not be used in the plural. Exceptions will be permitted for the use of plurals in abbreviations when the use of plurals clearly contributes to clarity and prevents ambiguities. 7. Use a hyphen between a number and a unit when used as an adjective. Example: 10- mL vial. 8. When the decimal point is used to express a fractional number, a 0 must be placed in front of it. Example: 0.28. 9. Do not use commas in long numerals. Either use no spaces or a full or half space every three digits. Example: 299792458 or 299 792 458, not 299,792,458. (This is an exception to APA style. In WordPerfect 5.1, a half space can be created using the Advance Right feature.) Alternatively, express large or small numbers in scientific notation. Example: 3.00 x 10°. Use the cross (WordPerfect character 6,39), not the letter x, in scientific notation. 10. Spell out numbers from one to nine except in a series of related numbers or where numerals are the standard usage. Example: 9 + 2. 11. Temperature symbols should include the value, then a space, then the unit. Example. 10 °C. 12. Latitude and longitude should have spaces between symbols and the following numbers. Example: 430° 02’ 22" N, 70° 42’ 57" W. 13. Centrifugation should be described in terms of relative centrifugal force, expressed in terms of g units of acceleration, rather than in terms of rotational speed in revolutions per minute. 14. Equations: Large, complex mathematical equations should be set off within the text where the use of exponents and built-up fractions is preferred for clarity. In WordPerfect 5.1, use the Equation Editor to create an equation of Paragraph type. 59 Example: Zou = = pAR? [sin@d® (1) Short mathematical expressions should be incorporated within the line of text and not displayed on a separate line. In WordPerfect 5.1, use the Equation Editor to create an equation of Character type. Avoid exponents having more than one level of characters. For example instead of ox?+y? , use exp(x’ + y’). Also, avoid the use of built-up fractions in the text. For example, instead of = , use 1/n or the negative exponent form n'. Likewise, avoid small-type expressions centered above or below arrows. Equations that are referred to later in the text should be numbered sequentially and referred to, for instance, as Equation (1). Avoid numbering equations that are not used. 15. If you are using WordPerfect 5.1, use the extended character set to create any symbols not found on the standard keyboard, such as Greek letters, mathematical symbols, etc. (This set is described in the document CHARACTR.DOC. To type characters in this set, press CTRL-V, then enter the two-number key for the character.) Otherwise, write in such symbols by hand. Accepted Manuscripts Copyright Forms If the manuscript is acceptable after refereeing, the author(s) will be asked to sign a copyright form, either transferring copyrights to the Washington Academy of Sciences or declaring that the paper is part of government work. The return of the signed form to the editors completes the acceptance process. Use Agreement Abstracting of material published in the Journal of the Washington Academy of Sciences is permitted with credit to the source. Instructors are permitted to photocopy isolated articles for noncommercial classroom use without fee. Proofs The corresponding author will receive a single set of page and illustration proofs. Proofs must be returned within 48 hours of their receipt. Delays in the return of the proofs can cause major delays in publication. All corrections should be marked clearly and directly on the page proofs; the authors are solely responsible for marking printer’s errors. Due to the high cost of corrections in proofs, no substantial alterations can be permitted after page proofs have been generated. Page Charges The Journal of the Washington Academy of Sciences depends in part on the payment of page charges to help defray the cost of publication. It is anticipated that the page charges of $50 60 (SRI en tees ae a er se per printed page ($40 for members of the Academy) will be paid by those authors who have funds available for that purpose, either from their institution or corporation or from the sponsor of the research. Members of the Academy in good standing without funds may apply through the editorial office for funds from the Past Presidents’ Fund to partially cover the cost of publication. However, editorial acceptance of a paper is not affected by payment or nonpayment and the payment of page charges is not a condition of publication. Reprints An order form for reprints will be sent with the page proofs. Reprints can be purchased, provided they are ordered when the corrected proofs are returned to the editors’ office. 61 é nie ary eee el ag | * 1 “i i y 7 ARC rd hy aaa le ay! rales x 4 4 ; sty - 4 - ae ct —— ar a RR os wore Te ey a 3 Korat * “YER Axi ex Ries eit ‘% a et og Pe sg 4; in = Heian vas A a % re} i aa Q sate att on w ‘: i a . ots OF fi) re bis é i? ir whee oi omy pa sina pe eS bn ‘i nes ae a eae, ey : ; : Dh. xe - ( y . REAL tiree See Branigl to Whiner, hy Bashantwe Cas Berets Sat on z wr sma Os our biad ons ind madera? = aNatiCN) ay Bs ib diity P a Me PRLS 3) pag ure ct ¥ besa = . be S : ‘ as ; ; A —— 7 - y 4 | Be wb: iy mn (ip hig gl Ged aks | 7 . ie Oe: 2 - a ; eas Ne ¢ , aloes wa ‘ sie wut ~~ ‘ f rely fee yy mies : 4 ) Pele P AAS aa? ad fe oe Dy t { ; : ; ‘ : ye } ' 2 *, j ? » ey 1 ie f 4 ? Ln. fel a} arin, 1 1¢ : ) Xie 5a) cit r , hs th v ‘ 4 . ‘ ' Oe Aiea Marseter.y. Torte a { fonds x i « i A eeu { ‘ y = > { 7 f j oe ‘ Hi { i 6 i i , RB ¢ u <= ty j ~ “ " i y ; - ‘ Ly oe ; ns +i ¥ ie ' - hi . a - : “ene . c a oh vw} (ve x md . ' Z rC { ‘ ‘ 7 re te a ‘ * 5 wae ? ) k \ é r PS cs { 2 is - i o ‘ } as ’ fe S i . } q : ; 5 § A e ; i} is ay es ~ - Cn A as 4) = \ yi 7 1 ; { 2 4 > a t x %. ~ * j + z ta) “ab ‘ - , 7 ye 7 3! 7 } = ‘ = 6 } { S ; ; 5 ; kK - > +“ a i> - c * ly a ; ( 4 a 4 shih a. x ne a < * ee 7] > avery eat) a ae ; rt fang melee ae a | MX ak oe uy _ cee Nate re = ; ike asvohaia oa AF wt ak abe hh : 7 botiarea , ye re on Stascaey Wane Section eG, 1 a ee “hae ia Dow Beit Goeinonts: ye Ay siak Wittens BS toes: Pipeirpabics f Ree 4 V ilaaees betel eae “ieee JA Ryineyr Ryo! Weochaiiaess: Ragiiessy. Want ae Satin hewn ee, Renee B Society of Wodbiigei Fereck : hk mete Poceses : tat Mixehibkhogy, % Srp ari #4 ps Wik: hr. =A) - tea } aR ei ice soly Bitton Tue: HaRKA, Vs Ladi inig ton Poa Se pete 2 bineegee! pal Cw i? Unica: PNuyt ioe al ¢ pital Seoviiie Ene . ia, wrt cé f, Wapeio Cheer coe alt |. a & Ot sia i Dees! Fee ar hy, Wirt ihaian Socih we a fj 2 hud or Aw ree: Loc Avs: es theew, m ft seri _ pan . ae, | biad Suet, Lx Chaves : ; . ‘ Tey ® soe Y vor Wu abeivagtione ‘. _ a 4 4 [eager c teasel’. Waahis gaps EIS: . | . Lute, Aon stad . y om ae mite, Wealimigion Sermon... 2.., ee a Soci “¥ Le a un ts w edb ene we xt i, Ww Zi ‘as re. f ¥ ’ tio “ty ; Ay Se Oral vot $c; a oA ¥ i Sule’ : =) ; Re eras). “ stion of Pipes joat cher, Chiwspeale Seteam - 3 Tot tig Aner nine ¥ ; xi ty Of Mart J hytdoh ists, Waahinyion Apee:Seas Operations %- sare? af Aa tragrimen! bi ' ogi =e ty at Ani one Washington Socthas ; ‘ a af Abin Hg, sdceaiies clive acct Pi fy Seite ‘ a : er .: ; ; ae A a iat abe eur) oe 3 _ iA F: “i at ‘7 oe | Asenesrbunii ei. Ait — a s ‘i> Ms ; co : ae } i ir ae reve a, é rT uv nae My, A : ane oD, glia d. ac ; = - 5 “ . ; . Phe Vip Be et AZ thee: a . » ; r : ; . ‘ 7 a , ¢ os c 7 ; oa 7B. ; i * HCkinexa! ois cs Wane i rit Pele Fs LUA: Ris! ae ee on 6h Restos iG eiiey, ‘Poknetowed Lingle” han Fietetns Sache, ‘Pipegurtin aes st inlonaie ee rig tad bili rv o Bok etal ak cat? oe ore Ds +) ere ‘ol re ? eee a AN cadences Phddnas i. Lahn iy aad “Kyte. Eusthiald. : velueaee tive ves ay “ae fost sine: A By Brawn, +Tal Neiaiat ‘atrd bert piste, DC" Se SOD. 5 ee fy a Fagegitt. Wnihingiow * cs ey ; ’ 7 = hy , 3 ee = DELEGATES TO THE WASHINGTON ACADEMY OF SCIENCES, REPRESENTING THE LOCAL AFFILIATED SOCIETIES Pmilosaphical Society of Washington .......:.6.62050.5). 1008604 s600s...: Lhomas R. Lettien | Patnropolocical Society Of Washington .. 52.2.0... cc. cc es eee ee ces Belford Lawson III | Bemarieal society Of WaSMINOTOM <). 5) acc. 8 ook Gace sowie solo Robie oe ak wakes os Kristian Fauchald | Mhemica society Of Washington 2... 5. ihe c bce ee ok wae eeeeok awe e ed Elise A. B. Brown Pnromolerical Society of Washington ....0...3.2. 5506060000 ceca ee ees F. Christian Thompson ena Mt COPEAD NIC SOCIETY) 4 niece cise oaks a/c ode ee bate Se clowns Sivatee beelels Stanley G. Leftwich | mreslgeical Society of Washington: 2... 2sccecc edie cece ceases eccctcewtacan James V. O'Connor | Medigansocicty Of the District of Columbia... .02 680586 obo. oe ce ces eel obs oe John P. Utz | Pistoneal Society of Washington, DCO . 2... o....sie. iw. ce eee ceed vale Thomas G. Manning | MCAS ORIELY Ol WASHINGTON 4.02. sents cee neces ctl eyes hele aade es hance Muriel Poston | Society of American Foresters, Washington Section .. 0.6.0. 00.5... e ce le lee Eldon W. Ross i! Pe PUNE SOCICLY Ol ENPIMCETS Wie. csc sales ons od nd sins vias sue Sele is ob ees se EE Alvin Reiner Institute of Electrical and Electronics Engineers, Washington Section ........ George Abraham i American Society of Mechanical Engineers, Washington Section ......... Clayton W. Robson ide Mminthological Society of Washington 2... ....0..2.....00. 00 ce cece eae Kendall G. Powers American Society for Microbiology, Washington Branch .................. Herman Schneider Society of American Military Engineers, Washington Post .................... James Donahue American Society of Civil Engineers, National Capital Section .............. John N. Hummel Society for Experimental Biology and Medicine, DC Section .............. Cyrus R. Creveling | Awemcrnational, Washington Chapter’ ..2/..6:....5 se. esc he dice ce eaes Pamela S. Patrick \ American Association of Dental Research, Washington Section ............. J. Terrell Hoffeld | American Institute of Aeronautics and Astronautics, National Capital SO MORN NE ep og so oc tny oats A LA ale PEt "eas ia olde, Suaiaet che le'siwids 6 04 Reginald C. Smith | Amenean Victeorological:Society, DC Chapter .........00 00005000000 eee ees A. James Wagner | Bast seietiec society Of Washington 2.2)... 002. hoe le edie de cect cee ee ee To be determined i Acoustical Society of America, Washington Chapter ......................., Richard K. Cook | Amchcanewucicar Society, Washington Section .........5..05 5... c cece ese c es ewes Kamal Araj } Institute of Food Technologists, Washington Section .................... George W. Irving, Jr. American Ceramic Society, Baltimore-Washington Section .................. Curtis A. Martin Pen MMC AES OGICDYA sdetails acon uc ck clad vk secu dina secs ede Paul Natishan | yy sminctonistony Of science Club)... foe ol cee ca cece ecw ee Albert G. Gluckman | American Association of Physics Teachers, Chesapeake Section ............. Robert A. Morse | Optical Society of America, National Capital Section ....................5- William R. Graver | American Society of Plant Physiologists, Washington Area Section ............. Steven J. Britz | Washington Operations Research/Management Science Council .............. John G. Honig Insthument Society of America, Washineton.Section ..............2...0...66. Donald M. Paul | American Institute of Mining, Metallurgical and Petroleum Engineers, RVI EMPTOR SEC UOM Emme ciate hla acl Wels ate oS ieee eae o snaei idles Lawl ss David M. Sutphin Neareonaln Capital ASTROMOMEES eck e soko ese ec ee cbc es ea tade see ces Robert H. McCracken Mathematics Association of America, MD-DC-VA Section ..................... Alice Schafer Disinctot Columbialnstitute Of Chemists ..).60)5 6. ee cc hd cee oe ee William E. Hanford District of Columbia Psychological Association ..... Re een eo Snr ae Sue Bogner J Masmimeton Paint; Fechnology Group: oF ./h edhe. ai cc sk a Yat lere sls die wow oc ee Lloyd M. Smith American Phytopathological Society, Potomac Division .................... Kenneth L. Deahl | Society for General Systems Research, Metropolitan Washington f CA OWEIE A een OE 2 Sikes 2 nn anne ee ANC ae RO RS John H. Proctor | human actors: society, Potomac Chapter). 2. 5.......65 605 feds bles cess Thomas B. Malone Amencan Fisheries Society, Potomac Chapter... /.... 0.5.2. 6...50.0 08 David A. Van Vorhees Association for Science, Technology and Innovation ........................ ... Ralph I. Cole | Basten Sociol osicaliSocietya wns... n0 ns ht so roe ee ee eee eee Ronald W. Manderscheid Institute of Electrical and Electronics Engineers, Northern Virginia ee Sell NO Daten tee ees taka eo to Se arts SU SINR, eka Pepe mane ea, Sea aR OE Blanchard D. Smith } Association for Computing Machinery, Washington Chapter ............. Charles E. Youman } Washinmeton! Statistical SOciebyi. ie oan oes ee ee lee mee ON, ciate Nancy Flournoy i Society of Manufacturing Engineers, Washington, DC Chapter ............... James E. Spates i Institute of Industrial Engineers, National Capital Chapter ................... James S. Powell | Delegates continue to represent their societies until new appointments are made. “ Washington Academy of Sciences 2nd Class Postage Paid 2100 Foxhall Rd. NW at Arlington, Va. Washinaton, DC 20007 ’ and additional mailing offices. Return Requested with Form 3579 Qa 1 a | 7 N H VOLUME 82 Number 2 Journal of the : Teboe WASHINGTON ACADEMY .. SCIENCES ISSN 0043-0439 Issued Quarterly at Washington, D.C. CONTENTS Articles: COLLIN C. CARBNO & SHIRLEY F. CARBNO, “Deviation Fractal Dimension” OS OOS OS PS SOONG SSS OA A OSes Oech Oe Cute) Cac Ont PCC Di Cnn rt ONO OG Ce Ch OO. CLIFFORD M. KROWNE & AARON P. KROWNE, “Nonlinear Dynamical Formulation for Describing Growth of Cancer Cells” .... 2.200.250. 2a. se ee bse 67 CARL M. HARRIS, EDWARD RATTNER & CLIFTON SUTTON, “How Senouslyis the US AIDS Population Understated?” . 2.0.0... eee Washington Academy of Sciences Founded in 1898 EXECUTIVE COMMITTEE President Stanley G. Leftwich President-Elect John H. Proctor Secretary Nina M. Roscher Treasurer Norman Doctor Past President Walter E. Bock Vice President, Membership Affairs Cyrus R. Creveling Vice President, Administrative Affairs Grover C. Sherlin Vice President, Junior Academy Affairs Marylin B. Krupsaw Vice President, Affiliate Affairs Thomas W. Doeppner Board of Managers James W. Harr John H. Proctor Herbert H. Fockler T. Dale Stewart William B. Taylor James H. Donahue REPRESENTATIVES FROM AFFILIATED SOCIETIES Delegates are listed on inside rear cover of each Journal. ACADEMY OFFICE 2100 Foxhall Road, N.W. Washington, D.C. 20007 Phone: (202) 337-2077 EDITORIAL BOARD Editor: Bruce F. Hill, Mount Vernon College Associate Editors: Milton P. Eisner, Mount Vernon Col- lege Albert G. Gluckman, University of Maryland Marc Rothenberg, Smithsonian Insti- tution Marc M. Sebrechts, Catholic Univer- sity of America Edward J. Wegman, George Mason University The Journal This journal, the official organ of the Washing- ton Academy of Sciences, publishes original scientific research, critical reviews, historical articles, proceedings of scholarly meetings of its affiliated societies, reports of the Academy, and other items of interest to Academy members. The Journal appears four times a year (March, June, September, and De- cember). The December issue contains a di- rectory of the current membership of the Academy. Subscription Rates Members, fellows, and life members in good standing receive the Journal without charge. Subscriptions are available on a calendar year basis, payable in advance. Payment must be made in U.S. currency at the following rates: U.S; and Canada *. . ..... See $25.00 Other countries | .....4.2. 73 30.00 Single copies, when available ....... 10.00 Claims for Missing Issues Claims will not be allowed if received more than 60 days after the day of mailing plus time normally required for postal delivery and claim. No claims will be allowed because of failure to notify the Academy of a change of address. Notification of Change of Address Address changes should be sent promptly to the Academy Office. Such notification should show both old and new addresses and zip codes. _ POSTMASTER: Send address changes to Washington Academy of Sciences, 2100 Fox- hall Road, N.W. Washington, DC 20007- 1199. Journal of the Washington Academy of Sciences (ISSN 0043-0439) Published quarterly in March, June, September, and December of each year by the Washing- ton Academy of Sciences, 2100 Foxhall Road, N.W., Washington, DC, 20007-1199. Second Class postage paid at Washington, DC and additional mailing offices. = Journal of the Washington Academy of Sciences, Volume 82, Number 2, Pages 63-66, June 1992 Deviation Fractal Dimension Collin C. Carbno Saskatchewan Telecommunications Shirley F. Carbno SAY-C Software There are many definitions (and estimates) of the fractal dimension of an object. Most of these definitions are equivalent in continuous domain, but some differ on discretized and digitized data (Barnsley, 1988; Dubic et al., 1987; Feder, 1988). As many researchers have discovered, the standard algorithms are somewhat awkward when it comes to processing actual experimental data (Knuth, 1992). We proposed another fractal dimension definition that for the case of an equally spaced time data series is conceptually simple. Given that one has a set of date points x', i = 1 to N, which were captured at equal time intervals one can form the point-to-point-variance (PTPV) s’, calcu- lated with the formula Spe) Ge Xe) Lal = N)/(N — A) where we adopt Iverson’s convention that [expression] has value | if expression is true, value 0 if expression is false (Moon, 1987). This concept is similar to the usual statistical variance, except that instead of the mean it uses the value of the ‘previous data point. From the PTPV one can form a point-to-point-deviation (PTPD) s by taking the square root of PTPV. The fractal dimension of an object is related to how that object behaves under scaling of some parameter. For example, in the a coastline case we have that L, the length of coastline, an island can be expressed as, Ken where r is length of the ruler used to measure the coastline, D is the fractal dimension, and K is a constant. In this case, the scaling parameter is the length 63 64 CARBNO AND CARBNO of the ruler used to measure the coastline. If the coastline has fractal dimension greater than one it follows that the length of the coastline becomes infinite as the ruler decreases in length. Mandelbrot’s insight was that over many scales of magnitude the value D typically remains a constant, so that fractal dimension is a good characterization of the scaling nature of a phenomenon. Using a similar approach, one asks how the PTPD will scale as a function of the time interval between data points. Now if the data behaved as if it were a straight line between the points, it follows that if you halve the time separation, that the PTPD will also halved. On the other hand, if each data points value is completely independent of previous data points, such as series formed from a set of dice throws, the value of PTPD will remain a constant. That is, if we throw the dice twice as often, we wouldn’t expect that the new intermediate values would reflect the dice’s previous value. That being the case we would expect that the PTPD for a series of dice throws to remain a constant. : These cases suggest a fractal dimension relationship SS hy where w is interval time separation between points, and K is a constant, D is fractal dimension, s = PTPD. In the limit as w approaches zero, if D approaches a constant value, then this value can be called the deviation fractal dimension of the data series. Typically, in the proposed method, one approximates the fractal dimension by comparing the PTPD calculated from data points two units apart. (Naturally, one can obtain whole series of variance values by using points in pairs that are three, four and so on time units apart.) In the case where the data values are independent, we would expect that D = 2, and in the case where the data values are smoothly changing in time, we would expect that D — 1. Thus, the deviation fractal is similar to other fractal dimensions and is related to the space-filling capability of the curve from which the data values are taken. Similarity of the PTPV to the usual variance of statistics may suggest to researchers ways of determining the statistical significance of particular devia- tion fractal diniension for a given data set. As such, investigation of this statistics may lead to a further synthesis of statistical and fractal dimension concepts. Past research into the connection between statistics and fractal has been fruitful with concepts such as fractal brownian functions, fractal brownian motion, R/S anal- ysis, and fractal statistics (Vicsek, 1992). As atest of the method, two sets of data were obtained. One dataset consisted of the minute-by-minute measurements of RM-60 Radiation Monitor (product of Aware Electronics) which provided via software connection to a personal computer minute by minute readings in miroroentgens per hour. A number of DEVIATION FRACTAL DIMENSION 65 data samples were obtained including runs of up to 48 hours periods. The data was loaded into a spreadsheet and the PTPD was calculated using points in pairs 2, 3,4, - - - 40 time units apart. The PT PV from this procedure remained fairly constant with some fluctuations as expected. Below is a table calculated from a 514 minute radiation file. Minute separation PTPD 1 minute 5.87 2 minutes 5.91 3 minutes 5.92 4 minutes SeD 5 minutes 5.84 Other data samples showed different patterns of PTPD so it appears that PTPD ‘1s basically a constant, and that the radiation received in any minute has no predictive power for radiation that will be received in the next minute. The second set of data consisted of daily weather temperatures (highs and lows) over a 27 year period. When we examined this data we found a PTPV for 1 to 10 day separation as follows (Weather Data provided by B. Farrer). PTPYV Fractal Dimension Sep. Highs Lows Data Pairs Points High Temp Low Temp l 139.2 106.7 9834 1.41 1.44 2 209.0 igen 9833 hy, 1.56 3 248.5 188.8 9832 1.66 1.58 4 273.8 212.8 9831 1:65 1.65 5 296.2 229.8 9830 1.79 1.66 6 307.8 244.6 9829 [e735 7 i 321.5 2535 9828 1c29 1.80 8 330.3 260.5 9827 1.84 1.78 9 336.5 2OT5 9826 1.79 1.64 10 344.2 277.9 9825 The fractal deviation dimension in the table were calculated using the formula D = 2 — In (s,/s,)/In (w./w,) The fractal value of 1.4 for one day separation suggests that the next day’s weather is somewhat dependent on today’s weather. A value of 1.4 indicates roughly that 60% of the next day’s weather is explained by today’s weather. At the eight day separations point, the variance is not dropping as fast as the time separation interval which can be interpreted as saying roughly that weather 7 days from now and 8 days from now are not that dependent on each other. 66 CARBNO AND CARBNO The PTPV were also calculated (using C) all the way from w = | day to w = 26 years. The PTPV showed as expected minimums on each yearly boundary. The yearly minimums themselves also showed a 4 year cycle pattern which is nicely explained by leap year phenomena. Does the deviation fractal dimension yield in the limit as w > O the same fractal dimension as other standard definition? As yet, the author hasn’t found a proof or a counter example for this conjecture. The proposed definition has some resemblance to the variational method of Dubic, being roughly based on the concept that a curve with fractal dimension is not differentiable. Thus any value which measures the rate of growth of the local derivative should be related to the fractal dimension. This suggests that perhaps a proof similar to what was used in the variation method of Dubic might work. Some numerical studies were done. Curves were generated with the fractal interpretation method of Michael Barnsley (Feder, 1988). The known fractal dimension of these curves was then compared with the calculations gave results that showed a strong correlation with known fractal dimensions. The differences most of the time ranged from 0 to around 0.15. The deviation fractal dimension was closer in the case of curves with higher fractal values and consistently gave high results for curves with fractal dimensions close to 1. Other computer trials were run in which the data points were brought closer and closer together. As expected the deviation fractal dimension showed a convergence towards the actual values. This suggests that the deviation fractal dimension is indeed a new way of obtaining fractal dimension of data. References Barnsley, M. (1988). Fractals Everywhere. Boston: Academic Press. Dubic, B. et al. (1987). Visual Communications and Image Processing IT, 845:241-248. Feder, J. (1988). Fractals. New York: Plenum. Knuth, D. (1992). Two Notes on Notation. The American Mathematical Monthly. 99:403-421. Moon, F. (1987). Chaotic Vibrations. New York: John Wiley & Sons. Vicsek, Tamas. (1992). Fractal Growth Phenomena. London: World Scientific. Weather Data—compliments of Bruce Farrer, Atmospheric Environment Service Station 401-6322-683 6620C, Qu’Appelle #1, Saskatchewan, Canada. Journal of the Washington Academy of Sciences, Volume 82, Number 2, Pages 67-78, June 1992 Nonlinear Dynamical Formulation for Describing Growth of Cancer Cells Clifford M. Krowne and Aaron P. Krowne 3810 Maryland Street Alexandria, VA 22309 ABSTRACT A set of equations characterizing the interactions between RNA, DNA and proteins is postulated to describe the growth of tumor cells. From this set of equations, a method to determine the fixed points of the system is presented including the use of the Jacobian matrix. Assessment of the nonlinear dynamics around these fixed points is provided. Introduction It is widely recognized that the common and perhaps only methods to treat cancer are radiation treatment (Hall), chemotherapy, surgery, microwave hy- perthermia (Carr, 1991), and combinations of these. All of these modalities have relative merits and disadvantages. But an overriding concern is that they all fail to reduce the number of tumor cells below a level at which they cease to replicate. Thus, enough tumor cells seem always to be left to form the starting population for another nonlinear growth process. There are hundreds of types of cancer (Golumbek et al., 1991) in a dozen general categories, including lym- | phoma, mammary (Friedenthal et al., 1984), prostate (Mendecki et al., 1980), myeloma, brain (Borok et al., 1988), lung, leukemia, and melanoma. The au- thor is particularly disturbed by the intractable nature of some of these cancers in terms of treating them, preventing them, and explaining why they are on the increase. Breast cancer (Rosen, 1991; Wallis, 1991) is one of the most publi- cized cancers and one of the most difficult to understand and treat effectively. In contrast, lymphoma may be treated relatively effectively by surgical removal, followed by a secondary modality if warranted (ie.; chemo- or radiation therapy). To both understand cancer cellular growth and prevent it in humans, it is 67 |) 68 KROWNE AND KROWNE | necessary to have a comprehensive research and clinical program using the latest developments and breakthroughs in fields which appear at first to appear wholly unrelated. These fields are microbiology, DNA engineering, molecular structure and crystallography, metabolic biology (May, 1987), nonlinear mathe- matics-physics-engineering, and clinical oncology. It is the intent of this paper to outline qualitatively and quantitatively the possibilities of nonlinear mathematics (Mackey and Milton, 1990; West, 1990) in explaining cell growth from a point of view of physics (Tomsovic and Heller, 1991; Pecora and Carroll, 1991; Corcoran, 1991; Scholl and Hupper, 1991; Chi and Vanneste, 1990; Ditto et al., 1990) and engineering (Krowne and Skor- upka, 1992; Skorupka et al., 1991; Carroll and Pecora, 1991; Parker and Chua, 1987; Matsumoto et al., 1985; Chua, 1984). A specific set of equations charac- terizing the interactions between RNA, DNA and proteins is assumed .to de- scribe the growth of tumor cells. From this set of equations, a method to deter- mine the fixed points of the system is described including the use of the Jacobian matrix. Assessment of the nonlinear dynamics around these fixed points is provided. | Metabolic Cell Growth Behavior Increase in the number of living cells, malignant or normal, requires a num- ber of biological processes to happen between two successive cell divisions. Cell mass should approximately double, and the genetic material encoded in the cellular DNA must be provided to the two daughter cells. The time between cellular divisions, the period T, is often divided into three unequal intervals; the G, interval, the S interval, and the G,M interval. The G, interval involves cellular events which are preparatory to DNA production in the following S interval. Molecular events during the G, interval are not as well understood as those occurring inside the S and GM intervals. A number of models have been proposed which suggest that it is in the G, interval that the cell decides to synthesize DNA or enter a quiescent state. And it may be at a particular development level (call it a critical point) in the G, | interval that the decision whether or not to synthesize DNA occurs. The mecha- nism for DNA triggering may be delocalized in space with a stochastic flavor caused by random molecular collisions at numerous cellular sites. This may also suggest that the critical point is spread over the G, interval with a function describing that characteristic. The G, time interval T , is dependent upon the nutrient environment, includ- ing growth factors. Some research information indicates that the sequence of | DESCRIBING CANCER CELL GROWTH | 69 events in the G, interval are related to proto-oncogenes, units of genetic infor- mation that code for growth-factor-like proteins. Oncogenes are expressed one after another during the cell cycle. Expression of a gene begins with the produc- tion of a string of mRNA (messenger RNA). This suggests that protein genera- tion results from RNA presence. After the G, interval and before the G,M interval is the S interval during which DNA 1s actually produced. Some consider this the most vital part of cell life. The last interval, the G,M, constitutes the time span during which the cell prepares for cell division, doing this by separating the DNA into two identical copies and then physically splitting the cell. Sometimes the cell grows and divides into two unequal sized or constituent filled daughter cells. This is called asymmetric cytokinesis. Both RNA and pro- tein are unequally distributed among the daughter cells. Cells which have more RNA in the earlier part of the G, interval, which immediately follows cell division, traverse the cell cycle faster than those that inherited less RNA. It appears that RNA is unequally divided in a random way between daughter cells. Formulation of Nonlinear Cellular Growth Equations Following other work (Kimmel, 1987), it is reasonable as a first approxima- tion to describe the linear interactions between the concentrations of RNA, DNA, and protein by the autonomous-like equations (no explicit time depen- dence ) — = —a,,R(t) + a,,P(t) dP a = ayR(t) — ayP(t) 2) 0 (Wend Be ae [bir )- P00 0 BR ree 4D ae (3) 0 Tate Dae = L- Here R(t), D(t), and P(t) are respectively the RNA, DNA, and protein con- centrations inside the cell. These equations describe the concentration changes within the cell over time in relation to the total time T it takes for a cell mitosis or division leading to replication. The first part of the process, taking time T,, is an interval involving no DNA synthesis. The next interval of time, of length T,, involves DNA synthesis. The last interval again involves no DNA synthesis. 70 KROWNE AND KROWNE This DNA behavior reflects the widely accepted view of what occurs during cellular growth and mitosis. Mitosis is that process whereby the cell duplicates the genetic information to split or divide. The other cellular organelles, proteins, and RNA are also replicated. Coefficient a,, gives the reduction of RNA over time and a,, gives the increase in RNA due to the ambient protein concentration P(t). Coefficient a,, gives the increase in protein concentration due to the ambient RNA concentration R(t). Coefficient a,, gives the reduction of protein over time. The boundary conditions under which equations (1) through (3) are solved are RO) 5 (0). DO) ss (4) RG) — FEE) DG Et by hz 1S) Boundary conditions (4), (5) indicate that at time t = 0 the normalized concen- trations of RNA, DNA, and protein are unity, that is, there is one cell with its concommitment of constituents, including the required amounts of RNA, DNA, and protein. After the period T elapses, the cell has undergone exactly one cell division, doubling the quantity of constituents including its RNA, DNA, and protein content. More exactly stated, just before period T, the cell has twice as much RNA, DNA, and protein, which will be uséd to produce the two new cells originating from the single cell in the mitotic process. Equations (1)-—(3) may be written in a more compact and general form for T, <= Dio as d R(t) ain oo On) pu) di Pe |- | din dp | P(t) | (6) D(t) 0 a Be (0) D(t) where T, is the translation and subtraction constant operator TCO ae) tc) C7) We may wish to call this system semi-autonomous because of the time transla- tion behavior. Furthermore, the original set of equations (1 )-—(3) has the third equation giving dD(t)/dt with different expressions in time. This amounts to an explicit time form. Thus the original system is in reality nonautonomous. Only in the interval T, P) (—a,, =F 2c,R sti CP) 0 (23) at, Ox, to second order in the nonlinear terms. Let us evaluate Of; /0x, 2. ae ee As 2 se eT) Pio) BP Ts J f(x) = 0 ) Notice that equations (19) and (20) are independent of D, the DNA concen- tration. This presumably makes sense because DNA is finally constructed from protein P and RNA R. Thus for the system as posed, we really have a 2 X 2 sized problem coupled to a third nonlinear equation. The 2 X 2 Jacobian J for this system can be extracted from (23) as Gai a 2b,R tr loyal eg)! (a, SF 2b,R =e b,>P) 25 (ap + 2¢,R aiGph)e (4457-426, Re eee) >) J = Sofu(x) = Once the 2 X 2 system is solved, the nonlinear dynamical behavior of DNA can be found from (21). One could also suppose, since D = f;(P) in form, that P has terms in D. Maybe, even R has D terms. Finally, D may not be independent of R and D. Nonlinear behavior can be determined locally by finding the fixed points of the system, linearizing about those fixed points, and studying the stability char- acteristics of the particular system under consideration. A fixed point x* occurs at DESCRIBING CANCER CELL GROWTH 73 x =0 (26) so that (8) becomes f,(x) = 0 (27) Fixed points occur when the system variable motion is zero as (26) states. It is possible to have only part of the variable space motion zero, in which case (26) will only hold for those appropriate components. x, = 0 il eke = IN (28) where the system is N dimensional. Linearizing (8) about the fixed point x* gives f(x) = £,(x*) + (x eEXS POs Fe (29) * Of, (x) Ox x=x Truncating this equation after the second term and noticing that the coefficient in the second term is just the Jacobian matrix evaluated at the fixed point x*, J = Sh) (30) Ox aie (29) can be written as Eat (x io dix x) (311) Defining a new variable y in reference to the fixed point y = x — x* (32) noting that (27) holds, and that Wir (33) we find that y = Jy (34) Linear solution of (34) allows the determination of the types of stability behavior about the individual fixed points. All of the mathematics of linear matrix analysis can be brought to bear on the solution of (34). Consider only the two equation system first, namely, (16) and (17), (19) and (20), and (25). Then apparently x* = 0 (35) is seen to be a fixed point satisfying (27) by inspection of the first five terms of (19) and (20). By (25), the Jacobian about the fixed point x* = 0 is 74 KROWNE AND KROWNE pass eh) eS By “ltl aco =e ot (36) 12 22 Eigenvalues and eigenvectors can be found setting the ith eigenvector solution for y as Se are (a) where y; is the ith eigenvector of y, v, the nontime part of the eigenvector, and ), the time eigenvalue. Placing (37) into (34) gives the equation [AI — J]v, = 0 (38) where I is the identity matrix. A nontrivial solution to (38) only holds if det [AI — J] = 0 (39) Putting (36) into (39) yields (A; + a1) — a2) th —aj2 (A; + a9) Eigenvalue solutions \, to (40) are = —F(an + Ag) + stan + 9)? — 4(a4,ay. — 4281) ]'/* (41) Because R, P, and D are real physical quantities, a,,, a,,, a5, a), must be real also in order to make the rate of change of RNA, DNA, and protein real, as a cursory inspection of (16) and (17) indicates. Based upon the fact that there are two eigenvalue solutions X., let us delineate the possible states of the system. Define 1a Fe tea + 9) (42) (40) BS sl(an + Ayo)? — 4(a4,A2. — ay)” (43) where A is half the system trace and equivalently for B B = [A? — det J]'”” where the determinant of J is found from (36) so that (41) becomes A, =A+B (44) Define C= A? — det J (45) so that B can be rewritten as B=C!/ (46) There are three fundamental cases. One is when C>0 (47) and the eigenvalues are real, meaning that the possibilities available for the system are decay and growth. The second is when DESCRIBING CANCER CELL GROWTH 75 C-<0 (48) and the eigenvalues are complex, meaning that the system may be in decaying oscillation or growing oscillation. The last case is when Ca (49) when the system has two degenerate eigenvalues and the system is either in decay or growth. Consider case (47) first. If the determinant of the Jacobian for the system is det J >0 (50) then (47) is satisfied when |A| > det J >0O G1) This places the eigenvalues in the interval Qe = 2A (52) and tells us that the system must be unstable in time and has what is known as an unstable node (Scholl, 1987), 1.e., the system undergoes growth if A>0O (53) and the system phase space trajectories diverge from the fixed point. For the half trace of the system being A <0 (943) the eigenvalues exist in the interval >A > 2A (5) and the system must be stable in time and has what is known as a stable node, _ that is, the system undergoes decay and the system phase space trajectories converge on the fixed point. If the determinant of the Jacobian for the system is det J < 0 (56) then (47) is always satisfied and Ne Sa (57) for an unstable solution, and , <0 (58) for a stable solution, and the system has a saddle point, provided (53) holds. For the half trace of the system obeying (54), 76 KROWNE AND KROWNE Ni 2A. (59) for a stable solution, and | A, > 0 (60) for an unstable solution, and the system still has a saddle point. The second fundamental case occurs for (48) when the system is found to possess oscillations. Equation (48) comes about when (50) holds and A? < det J (61) leading to a system with a stable focus if the half trace A of the system obeys (54). The phase space trajectory of the system is a converging elliptical spiral. The system has an angular frequency equal to w = |Im[A,, A,]| = {det J — A?}!” (62) and a decay or damping factor equal to d= Re[\,,\,] =A (63) and the two complex conjugate eigenvalues A, and i, are such that Im[A,] = —Im[),] (64) Re[\,]=Re[\] (65) If the trace of the system is greater than zero obeying (53), the system has an unstable focus and the phase space trajectory of the system is a diverging ellipti- cal spiral. The third fundamental case occurs for (49) causing degenerate eigenvalues which are real and take the sign of the half trace of the system A. According to the sign of A, see (56), an unstable or stable system node occurs. Conclusion We advocate the importance of a more holistic approach to cancer research. We strongly suggest that the effort be restructured so that several critical fields simultaneously interact and collaborate in order to develop a much more com- prehensive program. These fields include microbiology, DNA engineering, mo- lecular structure and crystallography, metabolic biology, nonlinear mathemat- icS-physics-engineering, and clinical oncology. | Because the fundamental underlying growth behavior of tumor cells is non- linear, this paper focused on studying a particular set of equations describing the interactions of DNA, RNA, and proteins in the context ofa linear system and its generalization to a nonlinear system. We hope our work will lead to further DESCRIBING CANCER CELL GROWTH la research into nonlinear dynamics of cancer cellular growth. There have been many studies which have dealt with the growth of cancer cell population dy- namics from an intercellular point of view (Duchting and Vogelsaenger, 1981; Suh and Weiss, 1984; De Boer and Hogeweg, 1986; Sluyser and Hart, 1983; Blum, 1974; Kimmel and Axelrod, 1991; Marusic et al., 1991; Chover and King, 1985; Merrill, 1984; Xu and Ling, 1988; Duchting and Dehl, 1980; Stein and Stein, 1990; Diekmann et al., 1984; Lasota and Mackey, 1984), but little work has been done to date to derive models for intracellular behavior (Free- man and Wilson, 1990; Gallez, 1984; Brooks, 1977; Baserga, 1984; Riddle et al., 1979; Traganos et al., 1982; Binggeli and Weinstein, 1986) which describe tu- mor growth. Acknowledgements Arye Rosen, Thomas Jefferson University, Philadelphia, PA was particularly helpful in pointing out the place microwave hyperthermia plays, in relation to the suite of modalities available, in treating cancer. Dr. Donald McRae, Vincent T. Lombardi Cancer Research Center, Depart. of Radiation Medicine, George- town University Medical Center, provided much insight into the clinical aspects of cancer and radiation biology. Dr. Mark Esrick, Physics Dept., Georgetown University, provided a few relevant references on theoretical biology and en- gaged us in several interesting discussions on nonlinear dynamics and chaos in biological systems. References Baserga, R. (1984). Growth in size and cell DNA replication. Expert. Ce// Res, 151:1-5. Binggeli, R., & Weinstein, R. C. (1986). Membrane potentials and sodium channels: hypotheses for growth regulation and cancer formation based on changes in sodium channels and gap junctions. J Th Bio, 123:377-401. ; Blum, H. F. (1974). Uncertainty of growth of cell populations in cancer. J Th Bio, 46:143-166. Borok, T. L., Winter, A., Laing, J., Paglione, R., Sterzer, F., Sinclair, I., & Plafker, J. (1988). Microwave hyperthermia radiosensitized iridium-192 for recurrent brain malignancy. Medical Dosimetry, 13:29-36. Brooks, R. F. (1977). Continuous protein synthesis is required to maintain probability of entry into S phase. Cell, 12:311-317. Carr, K. L. (1991). Microwave heating in medicine. Microwave J, 34:26-36. Carroll, T. L., & Pecora, L. M. (1991). Synchronizing chaotic circuits. IEEE Trans. Circuits Systems, 38:453- 456. Chi, C. C., & Vanneste, C. (1990). Onset of chaos and current-voltage characteristics of rf-driven Josephson junctions in the low-frequency regime. Phys Rev B, 42:9875-9895. Chover, J., & King, J. H. (1985). The early growth of cancer. J Math Bio, 21:329-346. Chua, L. O. (1984). Nonlinear Circuits. IEEE Trans. Circuits Systems, 31:69-87. Corcoran, E. Written by staff of Scientific American. (1991). Ordering chaos. In: Science and Business, Scientific American, 265:96-98. De Boer, R. J., & Hogeweg, P. (1986). Interactions between macrophages and T-lymphocytes: tumor sneak- ing through to helper T cell dynamics. J Th Bio, 120:331-351. Diekmann, O., Heijmans, H. J. A. M., & Thieme, H. R. (1984). On the stability of cell size distribution. J Math Bio, 19:227-248. 78 KROWNE AND KROWNE Ditto, W. L., Spano, M. L., Savage, H. T., Rauseo, S. N., Heagy, J., & Ott, E. (1990). Experimental observation of a strange nonchaotic attractor. Phys Rev Letts, 65:533-536. Duchting, W., & Dehl, G. (1980). Spread of cancer cells in tissue: modeling and simulation. Int J Biomed Comput, 11:175-195. Duchting, W., & Vogelsaenger, Th. (1981). Three dimensional pattern generation applied to spheroidal growth in a nutrient medium. /nt J Bio-Medical Computing, 12:377-392. Freeman, R. A., & Wilson, J. D. (1990). A two-channel hypothesis for regulation of cell division and differen- tiation. J Th Bio, 146:303-315. Friedenthal, E., Mendecki, J., Botstein, C., Sterzér, F., & Paglione, R. (1984). Hyperthermia in the treatment of local recurrence of breast cancer. Microwave J, 27:275-282. Gallez, D. (19dd). Cell membranes after malignant transformation Part I: Dynamic stability at low surface tension. J Th Bio, 111:323-340. Golumbek, P. T., Lazenby, A. J., Levitsky, H. I., Jaffee, L. M., Karasuyama, H., Baker, M., & Pardoll, D. M. N. (1991). Treatment of established renal cancer by tumor cells engineered to secrete interleukin-4. Science, 254:713-716. Hall, E. J. (1987). Radiobiology For The Radiologist. J. B. Lippincott, Philadelphia, Penn. Kimmel, M. (1987). Metabolic events in the cell cycle of malignant and normal cells. In: Thompson, J. R., & Brown, B. W. (ed.), Cancer Modeling, Marcel Dekker, New York, New York 5:215-235. Kimmel, M., & Axelrod, D. E. (1991). Unequal cell division, growth regulation and colony size of mamma- lian cells: mathematical model and analysis of experimental data. J Th Bio, 153:157-180. Krowne, C. M., & Skorupka, C. W. (1992). Nonlinear dynamical behavior of fields and currents in TED’s at microwave frequencies. In: Symmetry, Scaling, and Groups Session. URSI Radio Science Meeting Dig, 164. Lasota, A., & Mackey, M. C. (1984). Globally asymptotic properties of proliferating cell populations. J Math Bio, 19:43-62. Mackey, M.C., & Milton, J. G. (1990). A deterministic approach to survival statistics. J Math Bio, 28:33-48. Marusic, M., Bajzer, Z., Freyer, J. P., & Vuk-Pavlovic, S. (1991). Modeling autostimulation of growth in multicellular tumor spheroids. /nt J Biomed Comput, 29:149-158. Matsumoto, T., Chua, L. O., & Komuro, M. (1985). The double scroll. IEEE Trans. Circuits Systems, 32:798- 818. May, R. M. (1987). Nonlinearities and complex behavior in simple ecological and epidemiological models. In: Koslow, S. H., Mandell, A. J., and Shlesinger, M. F. (ed. conf organ.), Perspectives In Biological Dynamics And Theoretical Medicine, Ann New York Acad Sc. :1-31. Mendecki, J., Friedenthal, E., Botstein, C., Paglione, R., Sterzer, F. (1980). Microwave applicators for localized hyperthermia treatment of cancer of the prostate. Int. J. Radiation Oncology Biol. Phys, 6:1583- 1588. Merrill, S. J. (1984). Stochastic models of tumor growth and the probability of elimination by cytotoxic cells. J Math Bio, 20:305-320. Parker, T. S., & Chua, L. O. (1987). Chaos: A tutorial for engineers. Proc IEEE, 75:982-1008. Pecora, L. M., & Carroll, T. L. (1991). Pseudoperiodic driving: eliminating multiple domains of attraction using chaos. Phys Rev Letts, 67:945-948. Riddle, V. G. H., Pardee, A. B., & Rossow, P. W. (1979). Growth control of normal and transformed cells, J Supramolecular Struct, 11:529-538. | Rosen, R. (1991). Breast Cancer: An ignored epidemic. U C Davis Magazine, 8:17. Scholl, E. (1987). Nonequilibrium Phase Transitions in Semiconductors. Springer-Verlag, Berlin. Scholl, E., & Hupper, G. M. (1991). Dynamic hall effect as a mechanism for self-sustained oscillations and chaos in semiconductors. Phys Rev Letts, 66:2372-2375. Skorupka, C. W., Krowne, C. M., & Pecora, L. M. (1991). Strange nonchaotic behavior in a microwave TED stimulation. /st Experimental Chaos Conf Dig, 69: Sluyser, M., & Hart, G. (1983). Calculation of percentage of hormone independent transplantable cells in experimental mammary tumors. J 7h Bio, 100:701-707. Stein, W. C., & Stein, A. D. (1990). Testing and characterizating the two-stage model of carcinogenesis for a wide range of human cancers. J Th Bio, 145:95-122. Suh, O., & Weiss, L. (1984). The development of a technique for the morphometric analysis of invasion in cancer. J Th Bio, 107:547-561. Tomsovic, S., & Heller, E. J. (1991). Semiclassical dynamics of chaotic motion: unexpected long-time accu- racy. Phys Rev Letts, 67:664—-667. Traganos, F., Darzynkiewics, Z., & Melamed, M. R. (1982). The ratio of RNA to total nucleic acid content as a quantitative measure of unbalanced cell growth. Cytometry, 2:212-218. Wallis, C. (1991). A puzzling plague. Time, 137:48-52. West, B. J. (1990). Fractal Physiology and Chaos In Medicine. Studies of Nonlinear Phenomena in Life Science. World Scientific, 1:288. Xu, X., & Ling, Y. (1988). A study on the expectational model for tumor growth. Int. J. Biomed., 22:135-141. }| Journal of the Washington Academy of Sciences, Volume 82, Number 2, Pages 79-109, June 1992 How Seriously is the US AIDS Population Understated? Carl M. Harris, Edward Rattner, Clifton Sutton Department of Operations Research and Applied Statistics, George Mason University, Fairfax, Virginia 22030 August 14, 1992 ABSTRACT The work documented in this paper is an attempt at a realistic assessment of the true flow of AIDS cases entering the health-care system of the United States. Understatement in the numbers of AIDS diagnoses clearly causes serious problems for policy makers and analysts trying to understand the scope of the HIV/AIDS pandemic. From the perspective of the modeler and data analyst, the major problems would appear to result from /ags in the reporting and nonresponse. Our focus is therefore on the development of approaches for dealing with these problems. The major source of input data for our analysis is the nation- wide database of the CDC. Despite the fact that the United States and the world are struggling to deal with HIV/ AIDS. Yet, to date, there has been little agreement on the level and the sweep of the epi- demic. We believe that much of this confusion results from the very nature of the disease, from how affected people and the medical community have been responding, and, most importantly, from the manner in which HIV/AIDS data are collected and become available to the public over time. Most specifically, all such HIV/AIDS forecasting systems must convert reported AIDS cases back in time into a flow of HIV cases using what is often called backcalculation or backcasting. The systematic understatement of AIDS cases by year of diagnosis, however, compromises the very quality of the main statistical input stream for the analysis systems. In order to use such data in a valid way, there must be some sort of rational means for correcting underestimates of data and trends caused by reporting delays. Furthermore, it is well ac- cepted that some data are likely never to be captured at all by the national reporting system, although the medical needs of this hidden subpopulation will still burden the health-care system. Our work, then, focuses on the development of rational procedures for analyzing lags in data reporting and the subsequent development of a suggested model for extrapolating current AIDS statistics into a more useful data series. This is accomplished by applying multiple regression and goodness-of-fit techniques to the CDC data. In addition, a subgroup analysis is performed to determine what effect the shifting composition of the HIV+ popula- tion might have on future trends. We also examine the related question of underreporting. Introduction There are already strong indications that the HIV/AIDS epidemic will de- velop steeper growth levels in the period 1992-94, and will continue such in- 79 80 HARRIS, RATTNER, & SUTTON creases for the subsequent several years. The CDC has projected 58,000 to 85,000 new cases of AIDS for 1992 (CDC Centers for Disease Control, 1990b; Centers for Disease Control, 1990c). There have also been, of course, numerous estimates of the total size of the HIV-positive population. For example, (Don- dero Jr. et al., 1987) provided an estimate in 1987 of 1 to 1.5 million. The CDC’s own estimates have been for approximately 750,000 by January 1986; approxi- mately 1 million by June/July 1989; and still 1 million by December 1991 (Centers for Disease Control, 1990b; Centers for Disease Control, 1990c; Centers for Disease Control, 1992a). Additional discussions on key CDC meth- odological issues are in (Brookmeyer and Gail, 1988; Centers for Disease Con- trol, 1989b; Centers for Disease Control, 1990a; and Centers for Disease Con- trol, 1992b). The original objective of our project beginning in 1988 was the development of an innovative approach to the problem of forecasting the pace of the HIV/ AIDS epidemic. In part, this stemmed from the Dondero et al. 1987 assessment which sought firmer estimates/forecasts of prevalence and incidence (Dondero Jr. et al., 1987) and successive parallel statements emerging from the 1988 Leesburg Workshop (Office of Science and Technology Policy, 1988), as well as from the more recent report of the October 1989 CDC Workshop (Centers for Disease Control, 1990c). However, all of these key conferences over the years have led neither to a consensus nor to a resolution of conflicting concepts among the participants of how to reconcile outstanding issues. Our forecasts and model development work are most completely documented in (Harris et al., 1992). No matter what approach seems in favor at any time, there are major method- ological problems still open. We do not think that any of those active in this problem area were surprised at the strong criticism by the GAO of the quality of the major forecasting systems in June 1989 (General Accounting Office, 1989), even if one might differ with some of the specifics in that report. Through all of this, it appears that there is a persistent problem of converting CDC’s reported data to a more accurate picture of the status and evolution of the epidemic. The most profound issues revolve around the well recognized problems of lags in reporting and complete nonresponse (see, for example, CDC (Centers for Dis- ease Control, 1990c), Table 4). For example, questions of estimating HIV prevalence and incidence from prevailing data sources were the subjects of (Heyward and Curran, 1988; Curran et al., 1988; and Trafford, 1988). In addition, an Israeli paper by (Siegman-Igra et al., 1988) revealed some universal aspects of HIV/AIDS compromising the quality of data extrapolation, such as the role of mobility in the spread of the disease and possible AIDS longevity variations between various risk groups. In our studies, the cohorts have been time-based, and the focus of our work has been on projecting a time series that would anticipate the historical data | US AIDS POPULATION . 81 series evolving for the epidemic over the coming decades. Our analysis begins from CDC reports of AIDS cases, from which we éstimate the sizes of emerging morbidity-cohorts for each of the years of the US experience with the epidemic, thus the name Cohort Cascade for this major part of our modeling system. From the estimated development of the disease, we follow the progress of each annual _ cohort through the range of years of interest, and then aggregate the individual cohorts into an annual dataset of the expected numbers in various stages of illness for each of the years to the turn of the century. We have already made some first attempts at critical data adjustments, largely based on preliminary analyses of lags in reporting and on nonresponse. It is precisely because of our evolving attention to these data issues that we have prepared this current paper. The Walter Reed Institute of Medicine has formulated a series of stages to describe the progression of the disease (Brodt et al., 1986; Cowell and Hoskins, 1987; Redfield et al., 1986; Society of Actuaries, 1989). The first stage (abbre- viated as HIV+) refers to the initial asymptomatic phase during most of which the patient is in apparent good health. This is followed by the LAS (Lymph Adenopathy Syndrome) stage which includes symptoms of lymph system dys- function, and then by the ARC (AIDS-Related Complex) stage which includes a specific set of additional ailments of greater severity. Finally, AIDS (Acquired ImmunoDeficiency Syndrome) is the terminal stage of HIV illness and is marked by an additional set of opportunistic diseases, one of which is associated with the patient’s death. Remember throughout this work the important dis- tinctions between onset of HIV-positivity and the ultimate transition into AIDS. Other staging classifications are in use, e.g., based on T4 lymphocyte counts. These staging categories have not yet played an important role in forecasting system development; instead, they serve the same functions that any categorical structure serves in statistical analysis and reporting, as entity identifiers. Nonreporting of AIDS cases, intentional or not, is a particularly important - problem in estimates and projections of the epidemic. The findings of a study on the increased level of reporting of pneumonia deaths among younger adults indicated that many AIDS cases have been erroneously attributed to pneumonia (Stoneburner et al., 1988; and New York Times, 1988). More on this matter later. Designing An HIV/AIDS Prevalence And Incidence Model An effective modeling system for understanding the societal impact of HIV/ AIDS should provide several types of projections: @ National estimates of HIV death by year and by risk category; 82 HARRIS, RATTNER, & SUTTON @ National estimates of HIV morbidity, detailed by year, by risk category, and by stage; and @ The national (and local) resources (dollars, hospital days, physician-times, etc.) re- quired by HIV demands, detailed by year, by risk category, and by stage. The essential components of the incidence and prevalence portion of this modeling system would consist of two major linked sequential parts: 1. an Infection Model, which would generate the estimated annual (newly-infected) cohort for each year, 1976 to the present and for several future years. Such a model would estimate what type of person (and how many of that type) would move from HIV-negative to HIV-positive during a given span of time (e.g., the number of male heterosexuals who acquired HIV in 1989; the number of female [VDUs who will acquire HIV in 1993; etc.) 2. a Progression Model for translating the time cohort estimated by the Infection Model and its disaggregated major risk groups into specific levels of morbidity and mortality over time. The Progression Model would use estimated experience proba- bilities to obtain a plausible description of an annual HIV+ cohort’s movement through the several morbidity stages. The driving element of such a structure would be the number of newly in- fected (HIV-positive) persons in a given year, the infectivity base. When a per- son contracts the virus, a biological clock affecting his or her immunity system begins to tick; while the incubation times are recognized to vary greatly among individuals, each infected person is assumed to pass through each of several stages that have been defined by medical researchers. The precise dataset for HIV/AIDS prevalence and incidence we have devel- oped required the following techniques and assumptions: 1. Estimating the expected size of the newly infected HIV population for each year from 1976-86 by backcalculation using reporting lag-correction factor (i.e., numer- ical adjustments for data bias due to severe lateness in case reporting), without which the size of the seropositive population (past and future) would be systemati- cally and greatly understated; 2. Projecting annual growth rates for the HIV-infected population for the years 1986 through 1995 by a log-linear smoothing; and 3. Application of a nonresponse or hidden-case adjustment factor of 25% (which can be readily changed to any higher or lower assumed nonresponse level). Our modeling system is an aggregated one. Ideally, however, each of the newly-infected cohorts should be amenable to disaggregation. Each year’s newly infected (time-) cohort contains an increasingly heterogeneous population. Ini- tially, the overwhelming majority of HIV/AIDS seropositive were male and gay. In recent years, IV Drug Users and heterosexuals have become larger propor- tions of the newly-reported AIDS cases. Increasingly, [VDUs have begun to exceed Gays/Bi-s among new AIDS cases being reported in several states. The HIV+ cohort for a given year, the general aggregate number of that year’s newly-infected HIV cases, is a major element in estimating stage incidence levels US AIDS POPULATION ; 83 in subsequent years. The desegregated risk groups, subcohorts, are an even more meaningful statistical entity, since they can be the basis for estimating shifts among key demographic groups (particularly the gays, [VDUs, and heterosex- uals), by ultimately using a more complete Markov chain model. Undetected compensating shifts within the aggregate HIV+ cohort can obscure important behavioral changes among the key risk groups. Infectivity Issues and Backcalculation Remember that there are no data for the HIV cohort sizes in 1976, 1977, and succeeding years. Our Cohort Cascade structure provides a feasible basis for estimating those cohorts, but it is just one of a number of potential approaches for approximating incidence and prevalence through the years. However, such modeling systems necessarily start from data inputs on AIDS onset. Our model, just as many others, recreate the flow of new HIV cases through a range of years by backcalculation, which employs an estimate of the natural history of the progression of HIV/AIDS working in reverse from estimated numbers of new AIDS cases. Movement through the various stages of HIV disease is assumed to be linearly proportional to stage-population size. In addition to the staged proportionality of the Cohort Cascade portion of our model, we assume that infection is spread at a (piecewise) constant rate. That is to say, if the cumulative total number of surviving infectees by the beginning of year t is represented by N(t) and the number of deaths in year t by D(t), then the net increase of cumulative infected people in year t would be Mt + 1)- MO = aMD -— DO (1) or NG) Sexe allIN(@) =D): (2) ' Henceforth, we say that a, is the growth rate or “multiplier” for year t; we do permit some of the {a,} to be equal for purposes of illustration. In the years, 1976 to 1988, starting with ‘“‘one” infected individual, an annual constant growth rate of about 2.5 would have produced a cumulative total of about 3.38 million infected persons by the start of 1988. But if the epidemic started among an ultra-high risk group within the homosexual community, then spread to the remaining (lesser-high-risk) homosexual males, the growth rate in the first few years must have been considerably higher (than 2.5) with an implied lower subsequent growth rate, particularly as the awareness of the cause and deadliness of the disease emerged. That set of facts and assumptions is the basis for a surrogate infection model. 84 HARRIS, RATTNER, & SUTTON The estimation of the growth of the infected population (the “‘survivors’’, or total of the subpopulations in each of the stages) is the vital estimating factor in all estimation of HIV/AIDS; the infection model (using the multiplier measure) is the pivotal input to the progression model. A model responsive to this sce- nario would start with a very high growth rate level, which would lessen year-by- year within the homosexual community, then track the passage of the epidemic into the heterosexual population to estimate the number of US HIV seroposi- tives. Our actual estimates are derived using linear programming, with a full discussion of the procedure found in (Harris et al., 1992). Estimation of HIV populations and time-cohorts is, therefore, anchored in two major factual sets of data, the progression factors and AIDS-onset data (AIDS-onset: the year that an HIV patient transitions into the AIDS stage). AIDS-onset data have been reported regularly by CDC in earlier years in the AIDS Weekly Surveillance Report (AWSR, Table G) and currently in the HIV/ AIDS (Monthly, now Quarterly) Surveillance Report (HSR), and the progres- sion factors can be estimated from the several cohort studies of seropositives (for example, see Cowell and Hoskins, 1987). Backcalculating Past Cohort Sizes from Annual New-AIDS Cases Reported How do these new-AIDS cases relate to infectivity? The key of our infection model are the values of the {a,}. Equations (1) and (2) may be manipulated to yield _Mt+1)-NO+ DO) Oy NO) (3) It is clear that if the values of the {N(t)} and the {D(t)} are known accurately, then the {a,} can be easily determined. However, there is even more uncertainty about the year-by-year size of the seropositive population than there is about the yearly counts of the number of people who have reached the AIDS stage. There- fore the above expression for a, is not all that useful. Since it is not feasible to determine the {a,} directly using the {N(t)}, is desirable to estimate the unknown values of the multipliers using the more reliable counts of the number of people with AIDS. This can be done by exploit- ing the fact that, according to our conjecture, the yearly number of new AIDS cases is related to the sizes of the HIV+ cohorts via the proposed progression model. One way of doing this applies AIDS-to-COHORT equations in a back- calculation mode. If AIDS numbers were perfectly reported and if the probability structure we US AIDS POPULATION 85 Table 1.—Infected-Population Multiplier Estimates 1976 1977 1978 1979 1980 1981 1982 a 1983 1984 1985 Peo 4.07 72 1.36 0.85 0.70 0.55 0.45 0.37 0.32 assume for disease progression were also precise, then we would be able to write down a simultaneous system of linear equations in yearly cohort sizes, with the reported numbers of new AIDS cases on the right-hand side. To get the exact coefficients of such linear equations, we would begin from an assumption that the typical HIV case was infected on a given day. Defining that event as the start of Year | (and conclusion of Year 0) of that individual’s HIV-infection, that person’s probability of AIDS-onset is set in Cohort Cascade as .022 by the end of Year 3. Stated in equivalent terms, according to the factors of the Frankfurt study, a cohort of 1000 is expected to have 22 members develop AIDS by the end of 36 months, 1.e., the end of Year 3. By the end of the next year (Year 4), Cohort Cascade estimates that an additional 109 members will display AIDS, leaving 869 members AIDS-free by the end of Year 4; and so on. Infection Multipliers Each sequence of new AIDS cases relative to the unfolding numbers of sero- conversions creates a set of infection multipliers. The pattern of these multi- pliers determines the estimated growth of the epidemic in the future, and, in- deed, a major test of any HIV/AIDS modeling system is its ability to estimate these numbers well. We display in Table | the multipliers generated by the use of our linear-programming model applied to the CDC dataset of October 1990. The first major observation from Table | is that the multiplier pattern has been monotone decreasing and convex since the early years of 1976 and 1977 when little was known about the disease. The pattern passed through the mid- 40% range in 1983, when IVDUs became a significantly larger element of the newly-infected population. Since the IVDU subpopulation has been observed to be the most difficult to reach and seemingly the least amenable to behavioral changes, the declining multiplier rate is believed to be in a slowing rate of decline for the years after 1985. The work of (May and Anderson, 1987) contains a related approach to the estimate of infection spread. They calculated a growth rate of 1.0 as the result of a serologically based approximate doubling of the epidemic in San Francisco and New York within 10-11 months over the period 1978-1980. The multi- pliers we have settled on for our work (by backcasting estimation) are approxi- 86 HARRIS, RATTNER, & SUTTON mately 1 for both 1979 and 1980. However, we do believe that the numbers had to be considerably higher (originally, likely 2—3) in the years 1976 and 1977 in order to boost the epidemic during that time (i.e., doubling was much quicker in the very early days). The year 1978 is a kind of boundary year (in the sense that prior years provided insufficient cases to permit statistical precision), and is often the base from which to calculate forward estimates of multipliers, as it is for May and Anderson’s approach. But, we repeat, the ability to estimate disease spread is very much dependent on the manipulation of current data sources and the resultant extrapolations that they require. Methodological and Validation Issues Caveats, Confounding, and Data Biases Through it all, it seems very clear that there are two major data issues which must be handled carefully when using AIDS reports, namely, lagged reporting of both cases and deaths, and nonresponse. For example, every month in 1991, the CDC reported as “new cases” not only 1991 AIDS cases but also cases diagnosed in 1990, in 1989, and earlier years. The most recent years are always the most underreported. One effect of this lagged reporting is to give an illusion of “‘trend- washout” in the uncorrected data. That is, the real trend level, which shows a consistent increase each year, 1s most significantly affected by a damping bias in the recent years. Nonreporting is recognized also as a significant bias, since HIV and AIDS infectees are stigmatized in life as well as death. Denials of AIDS during the final months of life characterized the cases of several notables in recent years. Report- ing in a number of cities reviewed by CDC indicates a higher-than-expected incidence of pneumonia deaths among younger adults, which implies that the opportunistic disease, and not AIDS, may be on the death report in many cases (Stoneburner et al., 1988; and New York Times, 1988). Suicides during the latter stages of HIV are another likely source of misclassi- fied HIV cases. Also, pre-AIDS deaths (e.g., ARC-related) would not appear in the AIDS reports, although they are HIV-related. There are also some confounding elements in analyzing time-based cohorts. One such problem is the mobility (both travel and migration) of seropositives across regional and national boundaries. The mobility of homosexuals has been documented (e.g., see Siegman-Igra et al., 1988) as speeding the dispersion of HIV throughout the world. In estimating the growth of the epidemic, how should one handle the travel of seropositive, that is those already infected, from one locale to another? If a homosexual who acquires HIV in San Francisco returns home to Kansas City, is he then part of Kansas City’s AIDS growth? US AIDS POPULATION 87 Alternatively, does that mean that San Francisco’s AIDS growth has declined slightly? The National Opinion Research Center (New York Times, 1989a) has concluded that there is a great underreporting of Midwest cases, and suggests that it is a combination of avoidance (primarily by affluent whites) and a misat- tribution of Midwestern cases to either coast. This indicates that state and re- gional estimates will be subject to greater relative error than national projec- tions. Another, subtler, bias or dual-confounding factor is the outcome of an ex- tended set of AIDS-determinant factors. These factors were supplemented in late 1987 and were scheduled again for additional factors in 1993. This has the effect of shifting the mean AIDS-onset “discovery” date to an earlier point, thus leading to two misperceptions, the mean AIDS incubation period will now appear to be shorter and the mean AIDS longevity will appear to be longer. While other causes may truly shorten the AIDS incubation mean (e.g., inclusion of other groups, women, children, etc.) or may truly increase AIDS longevity (e.g., AZT, earlier start of healthcare, etc.), the extended-set bias should be accounted for and factored out. The introduction of intervention therapy to the HIV/AIDS population brings a potential bias to the estimates and projections of this model. Cohort Cascade is based on distributions for stages and years which derive from the natural history of the HIV/AIDS virus. PreAIDS intervention lengthens the time-in-stage for the seropositive, particularly those in the ARC stage. Intervention therapy after the onset of AIDS has been suggested as deferring death for about seven months, from 14 months to 21 months (Lemp et al., 1990). There are some limitations on the impact of these treatments, however. Firstly, not all patients are able to tolerate the treatments. Secondly, cost and other considerations have make them available to only a fraction of those who could benefit from their application. Thirdly, since AZT on any significant scale was not made generally available until about 1987, the beneficial impact of ‘Intervention therapy would be applicable primarily to two subpopulations, the ARC and AIDS patients, respectively about 30% of the HIV/AIDS population. Assuming that tolerance and availability will reduce that to about 10% of the HIV/AIDS population, one means of minimizing the bias in using natural history distribution solely is to create alternative cohorts with modified ARC and AIDS stage distributions reflecting the new extended-life findings. Lag Correction We are certainly not the first research team to worry about making correc- tions in data before utilizing them in the exercise of a modeling system. The | | papers by (Karon et al., 1988; Karon et al., 1989) discussed several methods for adjusting for lags in reported information, the Karon et al. Table 1 in (Karon et al., 1989) depicted data for 1984-1988 by quarters before and after adjustment. We derived our first empirical formula for lag correction, namely, [1 + (1/m)]? (i.e., at a point in time m months after the date of diagnosis, the expected | number of cases to be logged for that month ultimately would be the current ) number being reported by the CDC multiplied by [1 + (1/m)]? (see Harris et al., | 1992). That is, data in the June 1991 CDC report for the half-year ending June 1990 would have a value of m = 12, while m = 18 in the December 1991 CDC report for the same June 1990 data. That is, m equals the number of months ina given CDC report since the half-year data were “closed-out.”’ We did estimates for this 1984-88 period; the Karon et al. table 1 data are comparable to ours. Table 1/Cohort Cascade comparisons are: for 1984 AIDS (6019/6058), 1985 AIDS (11285/11297), 1986 AIDS (18590/18416), and for 1987 AIDS (27694/27502). (Later in this paper, we offer a detailed analysis of a more complete analysis of lag correction and the resultant development of a new set of correction factors.) (Brookmeyer and Damiano, 1989) also developed a set of adjustment factors; when the three sets of factors are compared, you get (Karon et al./Cohort Cas- cade/Brookmeyer & Damiano) for 4-6 months old data (1.33/1.36/1.39), 7-9 months (1.24/1.23/1.22), 10-12 months (1.19/1.17/1.15), 13-15 months (1.16/ 1.14/1.11), 6-18 months (1.13/1.11/1.08), 19-21 months (1.11/1.10/1.07), and 22-24 months (1.10/1.08/1.05). We have used our empirical adjustment factor, [1 + (1/m)]?, applied the progression rates derived from the natural history data of the Frankfurt study and the Cowell and Hoskins paper to structure our set of simultaneous equa- tions, 1.e., our backcasting/backcalculation model. That model has produced estimates for each annual cohort-population of HIV-infectees. Each cohort’s distribution through the successive HIV stages in each of the years following infection is aggregated to develop the ultimate dataset, a tabulation of cases by stage for the years 1976-93. There is an extrapolation component in the method used for the years beyond 1986, but it is an extrapolation based on the annual infectivity rates (i.e., the multipliers), and not on the annual newly diagnosed AIDS cases. 7 Cohort Cascade’s (old) lag-correction formula (as discussed above), applied to all past reported cohorts of annual new AIDS cases, provides more accurate estimates of AIDS cohorts. The actual results of these sorts of calculations on past published (e.g., in July 1987 and November 1988) AIDS cohorts estimated that they would increase through time as shown in Table 2. 88 HARRIS, RATTNER, & SUTTON US AIDS POPULATION 89 Table 2.—Validation of Lag-Correction Projections Date Original First Year of Data CDC CohCasc 2/90 CDC Percentage Cohort Publ’d Data Lag-C Update Difference 1984 7/87 5619 5972 5837 2.31% 1985 7/87 9756 10743 10867 —1.14% 1986 7/87 13583 17277 17601 —1.84% 1987 11/88 23771 27502 25863 6.34% Further support for our preliminary approach was offered by noting how (in Table 3) our (old) factors compared to the delay experience expected by the CDC (Centers for Disease Control, 1989a, Table 4). However, it is, of course, quite possible that both CDC and Cohort Cascade are using the wrong factors. It has therefore been a major target of our current work to examine this whole matter much more carefully, and the results of our investigation are found below in the “Construction ” section. One of our findings is a revision of Cohort Cascade’s lag-correction factors to a new set, Lag-Correction II: 3.3, 2.4, 2.1, 1.6, 1.4, which would replace line 3 of Table 3. The significance of these new factors is that they suggest a far greater impact on estimates of incidence, prevalence and trends for HIV/AIDS. More on this latter issue will be discussed in the “Concluding Remarks” section. More On Backcalculation/Backcasting No matter what approach is used in a modeling system for calculating back- wards in time, it is necessary to establish a database of past populations from which further analysis can append subsequent HIV cohorts. It should be kept in mind that the progression factors of the Frankfurt study may in time be dis- placed by later studies of HIV natural histories. In such case, the new progres- sion factors would be used to generate new projections. The Frankfurt factors were derived from a gay population; it would be desirable for subsequent pro- gression factors to be subpopulation specific, e.g., a set of gay progression fac- tors, another for [VDUs, and another for heterosexuals. It should be kept in mind that although we are analyzing AIDS reporting issues applying some degree of mathematical sophistication, the definition of Table 3.—Comparison of Lag-Correction Factors (CDC vs. Cohort Cascade) Months after diagnosis: 1 22 3 6 12 (CDC) Approx. formula: 3) // Well 17, 1.4 1 (CohCasc) Lag-Correction I: 4.0 223 1.8 1.4 1.2 90 HARRIS, RATTNER, & SUTTON what constitutes an AIDS case is subject to medically subjective modifications imposed by administrative decisions (e.g., definitional changes to the criteria-set of indicator diseases or symptoms, particularly in the Fall of 1987). The 1987 changes introduced several additional AIDS-indicator criteria. The expected result of the enhanced criteria-set was a “blip’”’, or sudden in- crease in the trend, which could expect a counterbalancing dip subsequently as the cases which were generated by the several new criteria move up in time for an earlier diagnosis. That is, these cases would indeed have been in later time included as they met one of the pre-1987 criteria. But having moved up, they depleted the stream of the later cases, which would result in a sparsity of candi- date cases for later transition into the AIDS population. We do, however, believe that our analyses of lags and underreporting, as documented later in this paper, are compromised only to a very minor degree by these questions, primarily because these data discontinuities are smoothed out over time. Our linear-programming method for backcalculation is actually dual in func- tion: it establishes a past history of the HIV epidemic’s growth through 1986, and it sets a bases for projecting infectivity multipliers to the years 1987 through 1993 and beyond. The past time series of multipliers, 1976-86, leads us to a consideration of the pattern for 1986 and beyond. If the pattern of decline continues, then the eight years 1986-93 would have infectivity multipliers de- clining gradually to about 12%. If the pattern levels off, the multipliers would instead average about 20%. Since we know that IVDUs are becoming a larger segment of the new AIDS cases, we might expect the multipliers to level-out to about a 20% average rate currently and in the near-future. Most analysts appear to accept the argument for a continuing, variable decline in the annual, aggre- gated infectivity rate through 1993 and beyond; but there is no consensus as to expected rate. In the end, the reliability of agreed-upon lag-correction factors will likely give all of us a better picture of how the epidemic will truly spread through the rest of this decade, particularly in light of the increasing focus on ameliorative therapies. Construction Of A More Rigorous Lag Correction Function The distribution of lags between the time of the diagnosis of AIDS and the time at which the case is reported to the CDC can be used to obtain an estimate of the number of cases that will ultimately be attributed to a specific diagnosis date. For instance, suppose it is known that 40% of all cases which will eventu- ally be attributed to a given month of diagnosis have been logged at the CDC at the end of a two month period following the diagnosis month. Then, if N(2) US AIDS POPULATION 91 cases have been ascribed to the given month of diagnosis at the completion of the two month period following that month, the number of cases which wil] ultimately be attributed to the month is simply N(2)/0.40 = 2.50 - M(2). In general, if the proportion of cases which have been reported within m months following a given month is F(m), and if N(m) is the number of cases which are known at that time, then the total number of cases which will eventually be reported for that date is given by Noo) = Mm)/F(m). That is, the inverse of the cumulative distribution can be used as a multiplicative lag correction factor. , _ The preceding expression for N (00) is not useful unless F(m) is known, and unfortunately there is no way to directly observe or measure the distribution function since F(m) = Nm)/N(co) depends on the unknown value N(co). However, by noticing and exploiting some key patterns in the available data one can develop a plausible approxima- tion to F(m). In this section, we will describe a step-by-step procedure for con- structing an estimate of the distribution function for the reporting lags, and the formula obtained from the method will be compared to the (implicit) estimates of the lag distribution reported by the (Centers for Disease Control, 1989a), Table 4. Exploratory Analysis The raw data on which the following analysis is based consists of the lags ‘corresponding to all reported AIDS cases which have been attributed to the years 1982, 1983,. . ., 1988. A lag is taken to belong to the interval (0, 1] if the case was reported to the CDC in the same month it was diagnosed or in the following month. If a case isn’t logged at the CDC until the second full month following the month of diagnosis, the lag is ascribed to the interval (1,2], and in general a case which is reported in the kth full month following the diagnosis month is assigned to (k — 1,k]. The data utilized for our work are the number of observed lags corresponding to cases diagnosed in 1988 for the intervals (0,1], (1,2], . . . , (29, 30], the number of observed lags corresponding to cases diagnosed in 1987 for the inter- vals (0,1), (1,2],. . .,(41,42], and the number of observed lags corresponding to 92 HARRIS, RATTNER, & SUTTON Table 4.— Various Ratios of the Cumulative Distributions of Lags for 1982-88 Indicating the Shapes of the Conditional Distributions for Lags No Greater than 30 Condit. distr. 1982 1983 1984 1985 1986 1987 1988 F(6)/F(30) 0.769 0.834 0.819 0.788 0.721 0.711 0.742 F(12)/F(30) 0.907 0.926 0.925 0.892 0.817 0.849 0.851 F(18)/F(30) 0.965 0.963 0.967 0.936 0.896 0.923 0.922 F(24)/F(30) 0.985 0.984 0.984 0.964 0.961 0.965 0.969 cases diagnosed in 1986 for the intervals (0,1], (1,2],. . ., (53,54]. The counts for intervals up to and including (65,66] for 1985, (77, 78] for 1984, (89,90] for 1983, and (95,96] for 1982 were also used. The data came from the CDC data base as reported through the second quarter of 1991. Consequently, even though some lags for the 1988 cases are reported in intervals up to (41,42], only the figures for the intervals through (29,30] represent complete counts. (For exam- ple, a December 1988 diagnosis having a lag greater than 30 months would not have been included in the second quarter 1991 CDC data base.) Thus, here we will only make use of intervals which are complete, and we wili not employ projected counts for intervals to which additional cases may eventually be added. Even though the cumulative distribution for the lags cannot be directly ob- served, one can learn something about the general shape of the lag distribution by examining the conditional distribution of all observed lags which are less than or equal to a certain value. For instance, of all the lags under consideration which belong to intervals up to (29,30], roughly 60% of them lie in the lowest three intervals. This fact, along with other information about the general shape of the distribution which can be gleaned from table 4, indicates that the distribu- tion of the lags is highly skewed with the distribution mean and median both being relatively small compared to some of the larger lag values contained in the data. Table 4 displays various values of the cumulative distribution for lags belong- ing to intervals up to (29,30] for the cohort years 1982-1988. While there are some similarities amongst the displayed values, particularly when values from adjacent years are compared, it is clear that the distribution of lags is not the same for each of the years, since if that were the case then along each row the displayed values would be identical. Additional insight pertaining to the lag distributions can be obtained by look- ing at values of other ratios of the cumulative distributions. (Note that since _ Nom,)/N(oo) N(m )/N(co) doesn’t depend on the unknown value of N(co), one can easily obtain the values of various ratios of the distribution functions even though the function values F(m,)/F(m,) = Mm,)/N(m,) US AIDS POPULATION : 93 Table 5.—Various Ratios of the Cumulative Distributions of Lags for 1982-88 Ratio 1982 1983 1984 1985 1986 1987 1988 F(6)/F(12) 0.848 0.900 0.885 0.883 0.883 0.837 0.872 f(12)/f(18) 0.939 0.962 0.957 0.953 0.912 0.920 0.923 F(18)/F(24) 0.980 0.978 0.983 0.971 0.933 0.956 0.951 F(24)/F(30) 0.985 0.984 0.984 0.964 0.961 0.965 0.969 F(30)/F(36) 0.985 0.986 0.986 0973 0.978 0.976 F(36)/F(42) 0.992 0.988 0.984 0.977 0.983 0.985 F(42)/F(48) 0.991 0.989 0.983 0.986 0.986 F(48)/F(54) 0.980 0.982 0.984 0.988 0.990 F(54)/F(60) - 0.962 0.987 0.991 0.989 F(60)/F(66) 0.970 0.979 0.991 0.992 F(66)/F(72) 0.967 0.990 0.992 F(72)/F(78) 0.971 0.991 0.994 cannot be observed.) Table 5 contains the values of certain ratios of the cumula- tive distributions for the years 1982-1988. One can see that the values along any of the rows are not identical, which indicates that the lag distributions are not the same for all of the years. However, one can note that there is a degree of similar- ity between row values of adjacent years, and that in some cases groups of three or more values in a row appear to be clustered around a central value. For each year, the observed pattern of the distribution ratios can, for the most part, be closely matched by fitting a function of the form m+a Benin Song (4) Functions of the form l An) (5) 1 + a/m + B/m?* can also be fit to yield a good degree of agreement with the observed values. It should be noted that the preceding approximation formulas are very compatible ‘with the data in the upper tail of the cohort distributions, but they are not uniformly accurate and tend to be rather loose in their agreement with the observed data in the first dozen or so intervals. Fitting The Distributions And Estimating Cohort Sizes The next step in construction an approximation to the cumulative distribu- tion of the lags is to obtain estimates of N(co) for each of the cohort years. Such estimates can then be used with the observed counts for the completed intervals to arrive at estimates of the cumulative distributions for each of the cohort years. For the 1988 data, fits of (4) and (5) were obtained by estimating the unknown 94 HARRIS, RATTNER, & SUTTON parameter a and 6 using the method of minimum y,?. The intervals (0,12], (12,13], 13,14], 14,15],. . .,(29,30] were employed and conditional probabil- ities were used since we had no data from the extreme upper tail of the distribu- tion. The decision of not breaking the initial interval (0,12] up into smaller intervals was based on a desire to place more emphasis on the trend of observed values in the upper tail of the distribution. This is compatible with the goal of accurately estimating F(30) so that the value of N(co) for 1988 can be estimated with N(30)/F(30), where F(30) is the estimate of F(30) obtained using minimum x? estimates of a and @. Since the observed interval counts indicate that (4) and (5) will provide reasonable fits in the upper tail, but not necessarily in the lower portion of the distribution, utilizing the larger interval (0,12] will allow a good approximation to be found even though the fitted function may be somewhat incompatible with the data from the initial monthly counts. Fits of (4) and (5) were also performed for the 1985, 1986, and 1987 cohorts, in each case using (0,12] as the first interval and then using all of the complete monthly counts. For each of the years 1985-88, fits of (4) and (5) were also done using the interval sets (O, Ke BS CK Seeks Sali | CK lie KS NO per tee (Ko 2. Ke een and | (KK {12 | (KG aka (Kee Ke Oi eer (K — 2) Ke Wk meena. where K is the upper endpoint for the last completed interval. These alternate sets of intervals place increased importance on accurately capturing the trend of the last 18 interval counts and the last 12 interval counts, respectively. Of all the fitted functions obtained for each cohort year, it is difficult to determine which one is best. One criteria would be to perform a x” goodness-of- fit test on each of the fits for a cohort, and select as the best fit the one which is most compatible with null hypothesis that (4) or (5) is the proper distribution function. However, this scheme is not necessarily ideal since it tends to favor the fits based on fewer intervals, and such fits can be more highly influenced (in a bad way) by quirks in the data. For each year, the interval counts tend to decrease as one gets further out into the tail of the distribution, but the observed counts do not follow a strictly decreasing pattern. Since it is reasonable to believe that the simple function which will best capture the trend of the cumulative distribution and lead to the best approximation of N(0o) is a smoothly increasing function consistent with a decreasing sequence of interval counts, and since the fits of (4) and (5) generally US AIDS POPULATION 95 Table 6.—Estimated Number of Diagnosed AIDS Cases which will Eventually be Attributed to 1985-88 (obtained by fitting the data from the various years separately) Year N(co) 1985 12029 1986 20031 1987 29801 1988 37271 imply decreasing interval counts, a method which is more sensitive to quirks in the sequence of observed counts may not be as good as one which focuses on finding a function which provides a reasonable fit over a longer sequence of intervals. It should also be kept in mind that because we are dealing with such large sample sizes in our goodness-of-fit tests, the high power of the testing procedure will cause some strong rejections of the null hypothesis even though from a practical viewpoint the fit coincides with the observed data rather well. Another way of arriving at a guess as to which fit may be best is to exploit the fact that Table 5 suggests that the cumulative distribution ratios are quite similar for adjacent years. Therefore, since we can observe that for 1987 we have N(42) = 1.040-N(30) (and noting additionally that for 1986 we have N(42) = 1.041 - N(30)), it seems reasonable to require that F(42) be close to 1.040 - N(30) for the 1988 data. Thus, the fit which provides an estimate of the 1988 cohort’s growth from 30 months to 42 months which is most consistent with the ob- served growth for the 1987 cohort can, in one sense, be deemed to be the best estimating function for 1988’s distribution. The various candidates for 1987 can be judged on how consistently they match 1986’s observed growth from 42 months to 54 months, and other years can be handled in a similar manner. In every case it turns out that one of the fits based on the original set of intervals does as good as any fit based on the alternate sets of intervals which place more emphasis on the trend in the extreme upper tail of the distribution. To arrive at estimates of N(co) for the various years, we proceeded as follows. For each year, we found the fit that corresponded to the largest goodness-of-fit test P-value (which is indicative of the highest degree of compatibility with the null hypothesis) and obtained the associated estimate of N(co). We then found the fit that most closely matched the observed growth rate for the preceding year and obtained the associated estimate of N(oo), and then averaged the two esti- mates to obtain the values displayed in Table 6. Finding A Universal Fit For Recent Years The estimated cohort sizes from Table 6 can be combined with the observed interval counts to arrive at estimates of the cumulative distribution functions for 96 HARRIS, RATTNER, & SUTTON the years 1985-88. Some values for the cohort distributions are shown in Table 7 below. From Table 7, it can be seen that with the exception of the first eight or nine months the cumulative distributions of the lags appear to be very similar for the 1987 and 1988 cohorts. One can also observe other similarities between portions of the distributions, but in general no strong patterns linking the four estimated distributions seem to exist. Of course, inaccuracies in the estimated values of N(co) for the various years could make the estimated distributions appear less similar (or more similar) to one another than is warranted. That being the case, it seems reasonable to see if a single (but perhaps with more than one piece) cumulative distribution approximation function can be found that is consistent with a large portion of the 1985-1988 data. However, it should be kept in mind that the values displayed in Table 7 (which are not based on any estimated parameters) indicate that the cumulative distributions are different for the various years, and so there exists no single formula which will completely agree with all of the observed interval counts. To find a function which provides a decent fit to both the 1987 and 1988 data, one can first define F,(m) to be the sum of the N(m) values for 1987 and 1988 divided by the sum of the estimates of N(oo) for those two years. Then an appropriate function can be fit to the computed values of F,(m) using regression. It turns out that functions of the form F*(m) = 1.0/[1.0 + G(m)], where G(m) is of the form Or are capable of providing good fits. One way of fitting such functions is to employ a least squares regression to fit the polynomial G(m) to the values of [1.0/ F,(m)]-1.0. Fits of this type were done using both the second degree and third degree versions of G(m), and fits were also done using values of F;(m) in place of F,(m), where the values of F,(m) are obtained by dividing the sum of the N(m) values for 1986, 1987, and 1988 by the sum of the N(co) estimates for those three years. To insure a good fit to the tail of the distribution, the fits were done using only the values m = 12, 18, 24, 30, 36, and 42. (For m = 36 and m = 42, 1988 US AIDS POPULATION Ly) Table 7.—Estimated Cumulative Distributions of Lags for 1985-88 Based on the N(co) Values Displayed in Table 6 Cum. dist. 1985 - 1986 1987 1988 F(1) 0.304 0.281 0.201 0.251 F(2) 0.468 0.425 0.373 0.422 F(3) 0553 0.500 0.484 0.517 F(4) 0.604 0.551 0.541 0.574 F(5) 0.638 0.585 0.585 0.612 F(6) 0.665 0.613 0.618 0.642 F(7) 0.688 0.634 0.645 0.663 F(8) 0.705 0.649 0.668 0.682 F(9) 0.720 0.662 0.687 0.697 F(10) 0.734 0.673 0.708 0.712 F(11) 0.743 0.684 0.723 0.724 F(12) 0.753 0.695 0.738 0.736 F(18) 0.790 0.762 0.802 0.797 F(24) 0.814 0.817 0.839 0.839 F(30) 0.844 0.850 0.870 0.865 F(36) 0.868 0.870 0.891 F(42) 0.888 0.885 0.904 F(48) 0.901 0.898 F(54) 0.912 0.907 F(60) 0.922 F(66) 0.929 values were not available and so the formulas for F,(m) and F,(m) had to be adjusted appropriately.) All things considered, the F,(m) values seemed to produce the more desirable fits. As expected, the third degree version of G(m) resulted in a nicer pattern of residuals; however, the second degree version of G(m) produced a fit that was actually superior in other aspects of being consistent with the observed data. It turns out that 1.0 1.0 + 4.83/m + 1.42/m? — 63.6/m? ’ which is obtained by averaging the coefficients of the least squares fit of the second and third degree versions G(m) to the values of F;(m), is an estimate of the cumulative distribution which exhibits remarkable consistencies with the observed data. UB is Table 8.—Preliminary Estimates of AIDS Cases Which Will Eventually be Attributed to 1985-88 (ob- tained by fitting the aggregated data from 1986-88) Year N(co) 1985 11994 1986 19797 1987 30049 1988 37411 98 HARRIS, RATTNER, & SUTTON Table 9.—Estimated Values of the Cumulative Distributions of Lags for 1985-88 Based on the N(co) Values Displayed in Table 8, With the Estimate F* Obtained from Fits of the 1986-88 Data Cum. dist. 1985 1986 1987 1988 Estimate F(6) 0.667 0.620 0.613 0.639 0.645 BG2) 0.755 0.703 OW7S82 0.733 0.727 F(18) 0.792 0.771 0.796 0.794 0.793 F(24) 0.816 0.827 0.832 0.835 0.834 F(30) 0.847 0.861 0.862 0.862 0.862 F(36) 0.870 0.880 0.883 0.882 F(42) 0.890 0.896 0.897 0.897 F(48) 0.903 0.908 0.909 F(54) 0.915 0.918 0.918 F(60) 0.924 0.925 F(66) 0.932 0.932 The values of N(co) resulting from F*(m) are given in Table 8. These. values are quite similar to the previously obtained ones displayed in Table 6. However, they differ enough from the previous values to yield noticeably different esti- mates of the cumulative distributions for the 1985-88 cohorts as can be seen by comparing the values from Table 7 with the values in Table 9. Not only does this estimate of the cumulative distribution produce an in- crease degree of similarity amongst the four distributions, but it also yields ratio values that are in line with what one might expect from the values displayed in Table 5. This can be observed in Table 10, where the ratio values of the esti- mated distribution have been added to the information that was contained in Table 5 (and the ratios for 1982 have been dropped). Even though the estimate just considered provides a more or less adequate fit of the upper tails of the 1986-88 distributions, it isn’t all that accurate over the first twelve monthly intervals. To determine a function which provides a good fit to the lower portion of the cumulative distribution one can repeat the proce- Table 10.—Various Ratios of the Cumulative Distributions of Lags for 1983-88 and of the Estimated Distribution Function Ratio 1983 1984 1985 1986 1987 1988 Estimate F(6)/F(12) 0.900 0.885 0.883 0.883 0.837 0.872 0.877 F(12)/F(18) 0.962 0.957 0.953 0.912 0.920 0.923 0.917 F(18)/F(24) 0.978 0.983 0.971 0.933 0.956 0.951 0.950 F(24)/F(30) 0.984 0.984 0.964 0.961 0.965 0.969 0.968 F(30)/F(36) 0.986 0.986 0.973 0.978 0.976 0.977 F(36)/F(42) 0.988 0.984 0.977 0.983 0.985 0.983 F(42)/F(48) 0.989 0.983 0.986 0.986 0.987 F(48)/F(54) 0.982 0.984 0.988 0.990 0.990 F(54)/F(60) 0.987 0.991 0.989 0.992 F(60)/F(66) 0.979 0.991 0.992 0.993 F(66)/F(72) 0.990 0.992 0.994 F(72)/F(78) 0.994 0.995 US AIDS POPULATION 99 Table 11.—Comparison of Values of F,(M) and the Regression-Based Approximation m F,(m) Approximation 2 0.398 0.398 3 0.499 0.502 4 0.556 0.555 5 0.596 0.592 6 0.627 0.623 Th 0.651 0.649 8 0.671 0.672 9 0.688 0.692 10 0.705 0.709 11 0.719 0.725 dure employed to fit the upper tail, only now using values of F,(m) and F,(m) for m = 2,3,4. .., 10, 11. (The m = 1 values were deleted since including them substantially worsened the quality of the fit. It should also be noted that the F,(m) and F,(m) values used here are based on the new N(co) values reported in Table 8.) For this part of the distribution, using the F,(m) values along with the third degree version of G(m) seemed to provide the best fit. Table 11 compares the values of F,(m) with the following approximation formula obtained from the regression: 1.0 1.0 + 5.0074/m — 10.3954/m? + 12.8785/m?> Further Discussion Upon putting all of the pieces together, we propose the following approxima- tion formula for the cumulative distribution of lags: O27 | m= 1 ' F(m) =§ (1.0 + 5.0074/m — 10.3954/m? + 12.8785/m3)! 1