T HE USE OF KNOWLEDGE IN SOCIET Y By F. A. HAYEK* I Wh at is th e proble m w e w ish to solv e w h e n w e try to construct a rational e conomic orde r? On ce rtain familiar assumptions th e answ e r is simple e nough . If w e posse ss all th e re le v ant information, if w e can start out from a giv e n syste m of pre fe re nce s and if w e command comple te know le dge of av ailable me ans, th e proble m w h ich re mains is pure ly one of logic. T h at is, th e answ e r to th e que stion of w h at is th e be st use of th e av ailable me ans is implicit in our assumptions. T h e conditions w h ich th e solution of th is optimum proble m must satisfy h av e be e n fully w orke d out and can be state d be st in math e matical form: put at th e ir brie fe st, th e y are th at th e marginal rate s of substitution be tw e e n any tw o commoditie s or factors must be th e same in all th e ir diffe re nt use s. T h is, h ow e v e r, is e mph atically not th e e conomic proble m w h ich socie ty face s. And th e e conomic calculus w h ich w e h av e de v e lope d to solv e th is logical proble m, th ough an important ste p tow ard th e solu- tion of th e e conomic proble m of socie ty, doe s not ye t prov ide an answ e r to it. T h e re ason for th is is th at th e "data" from w h ich th e e conomic calculus starts are ne v e r for th e w h ole socie ty "giv e n" to a single mind w h ich could w ork out th e implications, and can ne v e r be so giv e n. T h e pe culiar ch aracte r of th e proble m of a rational e conomic orde r is de te rmine d pre cise ly by th e fact th at th e know le dge of th e circum- stance s of w h ich w e must make use ne v e r e xists in conce ntrate d or inte grate d form, but sole ly as th e dispe rse d bits of incomple te and fre que ntly contradictory know le dge w h ich all th e se parate indiv iduals posse ss. T h e e conomic proble m of socie ty is th us not me re ly a proble m * T h e auth or is T ooke profe ssor of political e conomy and statistics at th e Univ e rsity of London (London Sch ool of Economics and Political Scie nce ). 520 T HE AMER ICAN ECONOMIC R EVIEW [SEPT EMBER of h ow to allocate "giv e n" re source s-if ''giv e n' is take n to me an giv e n to a single mind w h ich de libe rate ly solv e s th e proble m se t by th e se "data." It is rath e r a proble m of h ow to se cure th e be st use of re source s know n to any of th e me mbe rs of socie ty, for e nds w h ose re lativ e importance only th e se indiv iduals know . Or, to put it brie fly, it is a proble m of th e utilization of know le dge not giv e n to anyone in its totality. T h is ch aracte r of th e fundame ntal proble m h as, I am afraid, be e n rath e r obscure d th an illuminate d by many of th e re ce nt re fine me nts of e conomic th e ory, particularly by many of th e use s made of math e - matics. T h ough th e proble m w ith w h ich I w ant primarily to de al in th is pape r is th e proble m of a rational e conomic organization, I sh all in its course be le d again and again to point to its close conne ctions w ith ce rtain me th odological que stions. Many of th e points I w ish to make are inde e d conclusions tow ard w h ich div e rse path s of re asoning h av e une xpe cte dly conv e rge d. But as I now se e th e se proble ms, th is is no accide nt. It se e ms to me th at many of th e curre nt dispute s w ith re gard to both e conomic th e ory and e conomic policy h av e th e ir common origin in a misconce ption about th e nature of th e e conomic proble m of socie ty. T h is misconce ption in turn is due to an e rrone ous transfe r to social ph e nome na of th e h abits of th ough t w e h av e de v e lope d in de aling w ith th e ph e nome na of nature . II In ordinary language w e de scribe by th e w ord "planning" th e com- ple x of inte rre late d de cisions about th e allocation of our av ailable re source s. All e conomic activ ity is in th is se nse planning; and in any socie ty in w h ich many pe ople collaborate , th is planning, w h oe v e r doe s it, w ill in some me asure h av e to be base d on know le dge w h ich , in th e first instance , is not giv e n to th e planne r but to some body e lse , w h ich some h ow w ill h av e to be conv e ye d to th e planne r. T h e v arious w ays in w h ich th e know le dge on w h ich pe ople base th e ir plans is communicate d to th e m is th e crucial proble m for any th e ory e xplaining th e e conomic proce ss. And th e proble m of w h at is th e be st w ay of utilizing know le dge initially dispe rse d among all th e pe ople is at le ast one of th e main proble ms of e conomic policy-or of de signing an e fficie nt e conomic syste m. T h e answ e r to th is que stion is close ly conne cte d w ith th at oth e r que stion w h ich arise s h e re , th at of w h o is to do th e planning. It is about th is que stion th at all th e dispute about "e conomic planning" ce nte rs. T h is is not a dispute about w h e th e r planning is to be done or not. It is a dispute as to w h e th e r planning is to be done ce ntrally, by one auth ority for th e w h ole e conomic syste m, or is to be div ide d 1945] HAYEK: T HE USE OF KNOWLEDGE IN SOCIET Y 521 among many indiv iduals. Planning in th e spe cific se nse in w h ich th e te rm is use d in conte mporary controv e rsy ne ce ssarily me ans ce ntral planning-dire ction of th e w h ole e conomic syste m according to one unifie d plan. Compe tition, on th e oth e r h and, me ans de ce ntralize d planning by many se parate pe rsons. T h e h alf-w ay h ouse be tw e e n th e tw o, about w h ich many pe ople talk but w h ich fe w like w h e n th e y se e it, is th e de le gation of planning to organize d industrie s, or, in oth e r w ords, monopoly. Wh ich of th e se syste ms is like ly to be more e fficie nt de pe nds mainly on th e que stion unde r w h ich of th e m w e can e xpe ct th at fulle r use w ill be made of th e e xisting know le dge . And th is, in turn, de pe nds on w h e th e r w e are more like ly to succe e d in putting at th e disposal of a single ce ntral auth ority all th e know le dge w h ich ough t to be use d but w h ich is initially dispe rse d among many diffe re nt indiv iduals, or in conv e ying to th e indiv iduals such additional know le dge as th e y ne e d in orde r to e nable th e m to fit th e ir plans in w ith th ose of oth e rs. III It w ill at once be e v ide nt th at on th is point th e position w ill be diffe re nt w ith re spe ct to diffe re nt kinds of know le dge ; and th e answ e r to our que stion w ill th e re fore large ly turn on th e re lativ e importance of th e diffe re nt kinds of know le dge ; th ose more like ly to be at th e disposal of particular indiv iduals and th ose w h ich w e sh ould w ith gre ate r confide nce e xpe ct to find in th e posse ssion of an auth ority made up of suitably ch ose n e xpe rts. If it is today so w ide ly assume d th at th e latte r w ill be in a be tte r position, th is is be cause one kind of know le dge , name ly, scie ntific know le dge , occupie s now so promine nt a place in public imagination th at w e te nd to forge t th at it is not th e only kind th at is re le v ant. It may be admitte d th at, so far as scie ntific know le dge is conce rne d, a body of suitably ch ose n e xpe rts may be in th e be st position to command all th e be st know le dge av ailable th ough th is is of course me re ly sh ifting th e difficulty to th e proble m of se le cting th e e xpe rts. Wh at I w ish to point out is th at, e v e n assuming th at th is proble m can be re adily solv e d, it is only a small part of th e w ide r proble m. T oday it is almost h e re sy to sugge st th at scie ntific know le dge is not th e sum of all know le dge . But a little re fle ction w ill sh ow th at th e re is be yond que stion a body of v e ry important but unorganize d know l- e dge w h ich cannot possibly be calle d scie ntific in th e se nse of know l- e dge of ge ne ral rule s: th e know le dge of th e particular circumstance s of time and place . It is w ith re spe ct to th is th at practically e v e ry indiv idual h as some adv antage ov e r all oth e rs in th at h e posse sse s unique information of w h ich be ne ficial use migh t be made , but of 522 T HE AMER ICAN ECONOMIC R EVIEW [SEPT EMBER w h ich use can be made only if th e de cisions de pe nding on it are le ft to h im or are made w ith h is activ e coope ration. We ne e d to re me mbe r only h ow much w e h av e to le arn in any occupation afte r w e h av e comple te d our th e ore tical training, h ow big a part of our w orking life w e spe nd le arning particular jobs, and h ow v aluable an asse t in all w alks of life is know le dge of pe ople , of local conditions, and spe cial circumstance s. T o know of and put to use a mach ine not fully e mploye d, or some body's skill w h ich could be be tte r utilize d, or to be aw are of a surplus stock w h ich can be draw n upon during an inte rruption of supplie s, is socially quite as use ful as th e know le dge of be tte r alte rna- tiv e te ch nique s. And th e sh ippe r w h o e arns h is liv ing from using oth e rw ise e mpty or h alf-fille d journe ys of tramp-ste ame rs, or th e e state age nt w h ose w h ole know le dge is almost e xclusiv e ly one of te mporary opportunitie s, or th e arbitrage ur w h o gains from local diffe re nce s of commodity price s, are all pe rforming e mine ntly use ful functions base d on spe cial know le dge of circumstance s of th e fle e ting mome nt not know n to oth e rs. It is a curious fact th at th is sort of know le dge sh ould today be ge ne rally re garde d w ith a kind of conte mpt, and th at anyone w h o by such know le dge gains an adv antage ov e r some body be tte r e quippe d w ith th e ore tical or te ch nical know le dge is th ough t to h av e acte d almost disre putably. T o gain an adv antage from be tte r know le dge of facilitie s of communication or transport is some time s re garde d as almost dis- h one st, alth ough it is quite as important th at socie ty make use of th e be st opportunitie s in th is re spe ct as in using th e late st scie ntific discov e rie s. T h is pre judice h as in a conside rable me asure affe cte d th e attitude tow ard comme rce in ge ne ral compare d w ith th at tow ard pro- duction. Ev e n e conomists w h o re gard th e mse lv e s as de finite ly abov e th e crude mate rialist fallacie s of th e past constantly commit th e same mistake w h e re activ itie s dire cte d tow ard th e acquisition of such prac- tical know le dge are conce rne d-appare ntly be cause in th e ir sch e me of th ings all such know le dge is suppose d to be "giv e n." T h e common ide a now se e ms to be th at all such know le dge sh ould as a matte r of course be re adily at th e command of e v e rybody, and th e re proach of irra- tionality le v e le d against th e e xisting e conomic orde r is fre que ntly base d on th e fact th at it is not so av ailable . T h is v ie w disre gards th e fact th at th e me th od by w h ich such know le dge can be made as w ide ly av ailable as possible is pre cise ly th e proble m to w h ich w e h av e to find an answ e r. IV If it is fash ionable today to minimize th e importance of th e know l- e dge of th e particular circumstance s of time and place , th is is close ly conne cte d w ith th e smalle r importance w h ich is now attach e d to ch ange 1945] HAYEK: T HE USE OF KNOWLEDGE IN SOCIET Y 523 as such . Inde e d, th e re are fe w points on w h ich th e assumptions made (usually only implicitly) by th e "planne rs" diffe r from th ose of th e ir oppone nts as much as w ith re gard to th e significance and fre que ncy of ch ange s w h ich w ill make substantial alte rations of production plans ne ce ssary. Of course , if de taile d e conomic plans could be laid dow n for fairly long pe riods in adv ance and th e n close ly adh e re d to, so th at no furth e r e conomic de cisions of importance w ould be re quire d, th e task of draw ing up a compre h e nsiv e plan gov e rning all e conomic activ ity w ould appe ar much le ss formidable . It is, pe rh aps, w orth stre ssing th at e conomic proble ms arise alw ays and only in conse que nce of ch ange . So long as th ings continue as be fore , or at le ast as th e y w e re e xpe cte d to, th e re arise no ne w proble ms re quiring a de cision, no ne e d to form a ne w plan. T h e be lie f th at ch ange s, or at le ast day-to-day adjustme nts, h av e be come le ss im- portant in mode rn time s implie s th e conte ntion th at e conomic proble ms also h av e be come le ss important. T h is be lie f in th e de cre asing im- portance of ch ange is, for th at re ason, usually h e ld by th e same pe ople w h o argue th at th e importance of e conomic conside rations h as be e n driv e n into th e background by th e grow ing importance of te ch nological know le dge . Is it true th at, w ith th e e laborate apparatus of mode rn production, e conomic de cisions are re quire d only at long inte rv als, as w h e n a ne w factory is to be e re cte d or a ne w proce ss to be introduce d? Is it true th at, once a plant h as be e n built, th e re st is all more or le ss me ch anical, de te rmine d by th e ch aracte r of th e plant, and le av ing little to be ch ange d in adapting to th e e v e r-ch anging circumstance s of th e mome nt? T h e fairly w ide spre ad be lie f in th e affirmativ e is not, so far as I can asce rtain, borne out by th e practical e xpe rie nce of th e busine ss man. In a compe titiv e industry at any rate -and such an industry alone can se rv e as a te st-th e task of ke e ping cost from rising re quire s constant struggle , absorbing a gre at part of th e e ne rgy of th e manage r. How e asy it is for an ine fficie nt manage r to dissipate th e diffe re ntials on w h ich profitability re sts, and th at it is possible , w ith th e same te ch nical facilitie s, to produce w ith a gre at v arie ty of costs, are among th e commonplace s of busine ss e xpe rie nce w h ich do not se e m to be e qually familiar in th e study of th e e conomist. T h e v e ry stre ngth of th e de sire , constantly v oice d by produce rs and e ngine e rs, to be able to proce e d untramme le d by conside rations of mone y costs, is e loque nt te stimony to th e e xte nt to w h ich th e se factors e nte r into th e ir daily w ork. One re ason w h y e conomists are incre asingly apt to forge t about th e constant small ch ange s w h ich make up th e w h ole e conomic picture is probably th e ir grow ing pre occupation w ith statistical aggre gate s, w h ich 524 T HE AMER ICAN ECONOMIC R EVIEW [SEPT EMBER sh ow a v e ry much gre ate r stability th an th e mov e me nts of th e de tail. T h e comparativ e stability of th e aggre gate s cannot, h ow e v e r, be ac- counte d for-as th e statisticians se e m occasionally to be incline d to do-by th e "law of large numbe rs" or th e mutual compe nsation of random ch ange s. T h e numbe r of e le me nts w ith w h ich w e h av e to de al is not large e nough for such accide ntal force s to produce stability. T h e continuous flow of goods and se rv ice s is maintaine d by constant de - libe rate adjustme nts, by ne w dispositions made e v e ry day in th e ligh t of circumstance s not know n th e day be fore , by B ste pping in at once w h e n A fails to de liv e r. Ev e n th e large and h igh ly me ch anize d plant ke e ps going large ly be cause of an e nv ironme nt upon w h ich it can draw for all sorts of une xpe cte d ne e ds; tile s for its roof, statione ry for its forms, and all th e th ousand and one kinds of e quipme nt in w h ich it cannot be se lf-containe d and w h ich th e plans for th e ope ration of th e plant re quire to be re adily av ailable in th e marke t. T h is is, pe rh aps, also th e point w h e re I sh ould brie fly me ntion th e fact th at th e sort of know le dge w ith w h ich I h av e be e n conce rne d is know le dge of th e kind w h ich by its nature cannot e nte r into statistics and th e re fore cannot be conv e ye d to any ce ntral auth ority in statistical form. T h e statistics w h ich such a ce ntral auth ority w ould h av e to use w ould h av e to be arriv e d at pre cise ly by abstracting from minor diffe r- e nce s be tw e e n th e th ings, by lumping toge th e r, as re source s of one kind, ite ms w h ich diffe r as re gards location, quality, and oth e r particu- lars, in a w ay w h ich may be v e ry significant for th e spe cific de cision. It follow s from th is th at ce ntral planning base d on statistical informa- tion by its nature cannot take dire ct account of th e se circumstance s of time and place , and th at th e ce ntral planne r w ill h av e to find some w ay or oth e r in w h ich th e de cisions de pe nding on th e m can be le ft to th e "man on th e spot." V If w e can agre e th at th e e conomic proble m of socie ty is mainly one of rapid adaptation to ch ange s in th e particular circumstance s of time and place , it w ould se e m to follow th at th e ultimate de cisions must be le ft to th e pe ople w h o are familiar w ith th e se circumstance s, w h o know dire ctly of th e re le v ant ch ange s and of th e re source s imme diate ly av ailable to me e t th e m. We cannot e xpe ct th at th is proble m w ill be solv e d by first communicating all th is know le dge to a ce ntral board w h ich , afte r inte grating all know le dge , issue s its orde rs. We must solv e it by some form of de ce ntralization. But th is answ e rs only part of our proble m. We ne e d de ce ntralization be cause only th us can w e e nsure th at th e know le dge of th e particular circumstance s of time and place w ill be promptly use d. But th e "man on th e spot" cannot de cide 1945] IIAYEK: T HE USE OF KNOWLEDGE IN SOCIET Y 525 sole ly on th e basis of h is limite d but intimate know le dge of th e facts of h is imme diate surroundings. T h e re still re mains th e proble m of communicating to h im such furth e r information as h e ne e ds to fit h is de cisions into th e w h ole patte rn of ch ange s of th e large r e conomic syste m. How much know le dge doe s h e ne e d to do so succe ssfully? Wh ich of th e e v e nts w h ich h appe n be yond th e h orizon of h is imme diate know le dge are of re le v ance to h is imme diate de cision, and h ow much of th e m ne e d h e know ? T h e re is h ardly anyth ing th at h appe ns anyw h e re in th e w orld th at migh t not h av e an e ffe ct on th e de cision h e ough t to make . But h e ne e d not know of th e se e v e nts as such , nor of all th e ir e ffe cts. It doe s not matte r for h im w h y at th e particular mome nt more scre w s of one size th an of anoth e r are w ante d, w h y pape r bags are more re adily av ailable th an canv as bags, or w h y skille d labor, or particular mach ine tools, h av e for th e mome nt be come more difficult to acquire . All th at is significant for h im is h ow much more or le ss difficult to procure th e y h av e be come compare d w ith oth e r th ings w ith w h ich h e is also con- ce rne d, or h ow much more or le ss urge ntly w ante d are th e alte rnativ e th ings h e produce s or use s. It is alw ays a que stion of th e re lativ e importance of th e particular th ings w ith w h ich h e is conce rne d, and th e cause s w h ich alte r th e ir re lativ e importance are of no inte re st to h im be yond th e e ffe ct on th ose concre te th ings of h is ow n e nv ironme nt. It is in th is conne ction th at w h at I h av e calle d th e e conomic calculus prope r h e lps us, at le ast by analogy, to se e h ow th is proble m can be solv e d, and in fact is be ing solv e d, by th e price syste m. Ev e n th e single controlling mind, in posse ssion of all th e data for some small, se lf- containe d e conomic syste m, w ould not-e v e ry time some small adjust- me nt in th e allocation of re source s h ad to be made -go e xplicitly th rough all th e re lations be tw e e n e nds and me ans w h ich migh t possibly be affe cte d. It is inde e d th e gre at contribution of th e pure logic of ch oice th at it h as de monstrate d conclusiv e ly th at e v e n such a single mind could solv e th is kind of proble m only by constructing and constantly using rate s of e quiv ale nce (or "v alue s," or "marginal rate s of substitution"), i.e ., by attach ing to e ach kind of scarce re source a nume rical inde x w h ich cannot be de riv e d from any prope rty posse sse d by th at particular th ing, but w h ich re fle cts, or in w h ich is conde nse d, its significance in v ie w of th e w h ole me ans-e nd structure . In any small ch ange h e w ill h av e to conside r only th e se quantitativ e indice s (or "v alue s") in w h ich all th e re le v ant information is conce ntrate d; and by adjusting th e quantitie s one by one , h e can appropriate ly re arrange h is dispositions w ith out h av ing to solv e th e w h ole puzzle ab initio, or w ith out ne e ding at any stage to surv e y it at once in all its ramifications. 526 T HE AMER ICAN ECONOMIC R EVIEW [SEPT EMBER Fundame ntally, in a syste m w h e re th e know le dge of th e , re le v ant facts is dispe rse d among many pe ople , price s can act to co6rdinate th e se parate actions of diffe re nt pe ople in th e same w ay as subje ctiv e v alue s h e lp th e indiv idual to co6rdinate th e parts of h is plan. It is w orth conte mplating for a mome nt a v e ry simple and commonplace instance of th e action of th e price syste m to se e w h at pre cise ly it accomplish e s. Assume th at some w h e re in th e w orld a ne w opportunity for th e use of some raw mate rial, say tin, h as arise n, or th at one of th e source s of supply of tin h as be e n e liminate d. It doe s not matte r for our purpose -and it is v e ry significant th at it doe s not matte r- w h ich of th e se tw o cause s h as made tin more scarce . All th at th e use rs of tin ne e d to know is th at some of th e tin th e y use d to consume is now more profitably e mploye d e lse w h e re , and th at in conse que nce th e y must e conomize tin. T h e re is no ne e d for th e gre at majority of th e m e v e n to know w h e re th e more urge nt ne e d h as arise n, or in fav or of w h at oth e r ne e ds th e y ough t to h usband th e supply. If only some of th e m know dire ctly of th e ne w de mand, and sw itch re source s ov e r to it, and if th e pe ople w h o are aw are of th e ne w gap th us cre ate d in turn fill it from still oth e r source s, th e e ffe ct w ill rapidly spre ad th rough out th e w h ole e conomic syste m and influe nce not only all th e use s of tin, but also th ose of its substitute s and th e substitute s of th e se substitute s, th e supply of all th e th ings made of tin, and th e ir substitute s, and so on; and all th is w ith out th e gre at majority of th ose instrume ntal in bringing about th e se substitutions know ing anyth ing at all about th e original cause of th e se ch ange s. T h e w h ole acts as one marke t, not be cause any of its me mbe rs surv e y th e w h ole fie ld, but be cause th e ir limite d indiv idual fie lds of v ision sufficie ntly ov e rlap so th at th rough many inte rme diarie s th e re le v ant information is communicate d to all. T h e me re fact th at th e re is one price for any commodity-or rath e r th at local price s are conne cte d in a manne r de te rmine d by th e cost of transport, e tc.-brings about th e solution w h ich (it is just conce ptually possible ) migh t h av e be e n arriv e d at by one single mind posse ssing all th e information w h ich is in fact dispe rse d among all th e pe ople inv olv e d in th e proce ss. VI We must look at th e price syste m as such a me ch anism for com- municating information if w e w ant to unde rstand its re al function-a function w h ich , of course , it fulfills le ss pe rfe ctly as price s grow more rigid. (Ev e n w h e n quote d price s h av e be come quite rigid, h ow e v e r, th e force s w h ich w ould ope rate th rough ch ange s in price still ope rate to a conside rable e xte nt th rough ch ange s in th e oth e r te rms of th e contract.) T h e most significant fact about th is syste m is th e e conomy of know le dge 1945] HAYEK: T HE USE OF KNOWLEDGE IN SOCIET Y 52 7 w ith w h ich it ope rate s, or h ow little th e indiv idual participants ne e d to know in orde r to be able to take th e righ t action. In abbre v iate d form, by a kind of symbol, only th e most e sse ntial information is passe d on, and passe d on only to th ose conce rne d. It is more th an a me taph or to de scribe th e price syste m as a kind of mach ine ry for re giste ring ch ange , or a syste m of te le communications w h ich e nable s indiv idual produce rs to w atch me re ly th e mov e me nt of a fe w pointe rs, as an e ngine e r migh t w atch th e h ands of a fe w dials, in orde r to adjust th e ir activ itie s to ch ange s of w h ich th e y may ne v e r know more th an is re fle cte d in th e price mov e me nt. Of course , th e se adjustme nts are probably ne v e r "pe rfe ct" in th e se nse in w h ich th e e conomist conce iv e s of th e m in h is e quilibrium analysis. But I fe ar th at our th e ore tical h abits of approach ing th e proble m w ith th e assumption of more or le ss pe rfe ct know le dge on th e part of almost e v e ryone h as made us some w h at blind to th e true function of th e price me ch anism and le d us to apply rath e r misle ading standards in judging its e fficie ncy. T h e marv e l is th at in a case like th at of a scarcity of one raw mate rial, w ith out an orde r be ing issue d, w ith out more th an pe rh aps a h andful of pe ople know ing th e cause , te ns of th ousands of pe ople w h ose ide ntity could not be asce rtaine d by month s of inv e stigation, are made to use th e mate rial or its products more sparingly; i.e ., th e y mov e in th e righ t dire ction. T h is is e nough of a marv e l e v e n if, in a constantly ch anging w orld, not all w ill h it it off so pe rfe ctly th at th e ir profit rate s w ill alw ays be maintaine d at th e same constant or "normal" le v e l. I h av e de libe rate ly use d th e w ord "marv e l" to sh ock th e re ade r out of th e complace ncy w ith w h ich w e ofte n take th e w orking of th is me ch anism for grante d. I am conv ince d th at if it w e re th e re sult of de libe rate h uman de sign, and if th e pe ople guide d by th e price ch ange s unde rstood th at th e ir de cisions h av e significance far be yond th e ir imme diate aim, th is me ch anism w ould h av e be e n acclaime d as one of th e gre ate st triumph s of th e h uman mind. Its misfortune is th e double one th at it is not th e product of h uman de sign and th at th e pe ople guide d by it usually do not know w h y th e y are made to do w h at th e y do. But th ose w h o clamor for "conscious dire ction"-and w h o cannot be - lie v e th at anyth ing w h ich h as e v olv e d w ith out de sign (and e v e n w ith out our unde rstanding it) sh ould solv e proble ms w h ich w e sh ould not be able to solv e consciously-sh ould re me mbe r th is: T h e proble m is pre - cise ly h ow to e xte nd th e span of our utilization of re source s be yond th e span of th e control of any one mind; and, th e re fore , h ow to dispe nse w ith th e ne e d of conscious control and h ow to prov ide induce me nts w h ich w ill make th e indiv iduals do th e de sirable th ings w ith out anyone h av ing to te ll th e m w h at to do. 528 T HE AMER ICAN ECONOMIC R EVIEW [ SEPT EMBER T h e proble m w h ich w e me e t h e re is by no me ans pe culiar to e co- nomics but arise s in conne ction w ith ne arly all truly social ph e nome na, w ith language and most of our cultural inh e ritance , and constitute s re ally th e ce ntral th e ore tical proble m of all social scie nce . As Alfre d Wh ite h e ad h as said in anoth e r conne ction, "It is a profoundly e rrone ous truism, re pe ate d by all copy-books and by e mine nt pe ople w h e n th e y are making spe e ch e s, th at w e sh ould cultiv ate th e h abit of th inking w h at w e are doing. T h e pre cise opposite is th e case . Civ ilization ad- v ance s by e xte nding th e numbe r of important ope rations w h ich w e can pe rform w ith out th inking about th e m." T h is is of profound sig- nificance in th e social fie ld. We make constant use of formulas, symbols and rule s w h ose me aning w e do not unde rstand and th rough th e use of w h ich w e av ail ourse lv e s of th e assistance of know le dge w h ich indiv idually w e do not posse ss. We h av e de v e lope d th e se practice s and institutions by building upon h abits and institutions w h ich h av e prov e d succe ssful in th e ir ow n sph e re and w h ich h av e in turn be come th e foundation of th e civ ilization w e h av e built up. T h e price syste m is just one of th ose formations w h ich man h as le arne d to use (th ough h e is still v e ry far from h av ing le arne d to make th e be st use of it) afte r h e h ad stumble d upon it w ith out unde rstanding it. T h rough it not only a div ision of labor but also a cobrdinate d utiliza- tion of re source s base d on an e qually div ide d know le dge h as be come possible . T h e pe ople w h o like to de ride any sugge stion th at th is may be so usually distort th e argume nt by insinuating th at it asse rts th at by some miracle just th at sort of syste m h as spontane ously grow n up w h ich is be st suite d to mode rn civ ilization. It is th e oth e r w ay round: man h as be e n able to de v e lop th at div ision of labor on w h ich our civ ilization is base d be cause h e h appe ne d to stumble upon a me th od w h ich made it possible . Had h e not done so h e migh t still h av e de - v e lope d some oth e r, altoge th e r diffe re nt, type of civ ilization, some th ing like th e "state " of th e te rmite ants, or some oth e r altoge th e r un- imaginable type . All th at w e can say is th at nobody h as ye t succe e de d in de signing an alte rnativ e syste m in w h ich ce rtain fe ature s of th e e xisting one can be pre se rv e d w h ich are de ar e v e n to th ose w h o most v iole ntly assail it-such as particularly th e e xte nt to w h ich th e indi- v idual can ch oose h is pursuits and conse que ntly fre e ly use h is ow n know le dge and skill. VII It is in many w ays fortunate th at th e dispute about th e indispe nsa- bility of th e price syste m for any rational calculation in a comple x socie ty is now no longe r conducte d e ntire ly be tw e e n camps h olding diffe re nt political v ie w s. T h e th e sis th at w ith out th e price syste m w e 1945] HAYEK: T HE USE OF KNOWLEDGE IN SOCIET Y 529 could not pre se rv e a socie ty base d on such e xte nsiv e div ision of labor as ours w as gre e te d w ith a h ow l of de rision w h e n it w as first adv ance d by v on Mise s tw e nty-fiv e ye ars ago. T oday th e difficultie s w h ich some still find in acce pting it are no longe r mainly political, and th is make s for an atmosph e re much more conduciv e to re asonable discussion. Wh e n w e find Le on T rotsky arguing th at "e conomic accounting is unth inkable w ith out marke t re lations"; w h e n Profe ssor Oscar Lange promise s Profe ssor v on Mise s a statue in th e marble h alls of th e future Ce ntral Planning Board; and w h e n Profe ssor Abba P. Le rne r re - discov e rs Adam Smith and e mph asize s th at th e e sse ntial utility of th e price syste m consists in inducing th e indiv idual, w h ile se e king h is ow n inte re st, to do w h at is in th e ge ne ral inte re st, th e diffe re nce s can inde e d no longe r be ascribe d to political pre judice . T h e re maining disse nt se e ms cle arly to be due to pure ly inte lle ctual, and more particularly me th odological, diffe re nce s. A re ce nt state me nt by Profe ssor Jose ph Sch umpe te r in h is Capitalism, Socialism and De mocracy prov ide s a cle ar illustration of one of th e me th odological diffe re nce s w h ich I h av e in mind. Its auth or is pre - e mine nt among th ose e conomists w h o approach e conomic ph e nome na in th e ligh t of a ce rtain branch of positiv ism. T o h im th e se ph e nome na accordingly appe ar as obje ctiv e ly giv e n quantitie s of commoditie s impinging dire ctly upon e ach oth e r, almost, it w ould se e m, w ith out any inte rv e ntion of h uman minds. Only against th is background can I account for th e follow ing (to me startling) pronounce me nt. Profe ssor Sch umpe te r argue s th at th e possibility of a rational calculation in th e abse nce of marke ts for th e factors of production follow s for th e th e orist "from th e e le me ntary proposition th at consume rs in e v aluating ('de - manding') consume rs' goods ipso facto also e v aluate th e me ans of production w h ich e nte r into th e production of th e se goods."' T ake n lite rally, th is state me nt is simply untrue . T h e consume rs do noth ing of th e kind. Wh at Profe ssor Sch umpe te r's "ipso facto" pre - sumably me ans is th at th e v aluation of th e factors of production is 'J. Sch umpe te r, Capitalism, Socialism, and De mocracy (Ne w York, Harpe r, 1942), p. 175. Profe ssor Sch umpe te r is, I be lie v e , also th e original auth or of th e myth th at Pare to and Barone h av e "solv e d" th e proble m of socialist calculation. Wh at th e y, and many oth e rs, did w as me re ly to state th e conditions w h ich a rational allocation of re source s w ould h av e to satisfy, and to point out th at th e se w e re e sse ntially th e same as th e condi- tions of e quilibrium of a compe titiv e marke t. T h is is some th ing altoge th e r diffe re nt from sh ow ing h ow th e allocation of re source s satisfying th e se conditions can be found in prac- tice . Pare to h imse lf (from w h om Barone h as take n practically e v e ryth ing h e h as to say), far from claiming to h av e solv e d th e practical proble m, in fact e xplicitly de nie s th at it can be solv e d w ith out th e h e lp of th e marke t. Se e h is Manue l d'Jconomie pure (2nd e d., 1927), pp. 233-34. T h e re le v ant passage is quote d in an English translation at th e be gin- ning of my article on "Socialist Calculation: T h e Compe titiv e 'Solution,'" in Economica, Ne w Se rie s, Vol. VIII, No. 26 (May, 1940), p. 125. 530 T HE AMER ICAN ECONOMIC R EVIEW implie d in, or follow s ne ce ssarily from, th e v aluation of consume rs' goods. But th is, too, is not corre ct. Implication is a logical re lationsh ip w h ich can be me aningfully asse rte d only of propositions simultane ously pre se nt to one and th e same mind. It is e v ide nt, h ow e v e r, th at th e v alue s of th e factors of production do not de pe nd sole ly on th e v alua- tion of th e consume rs' goods but also on th e conditions of supply of th e v arious factors of production. Only to a mind to w h ich all th e se facts w e re simultane ously know n w ould th e answ e r ne ce ssarily follow from th e facts giv e n to it. T h e practical proble m, h ow e v e r, arise s pre - cise ly be cause th e se facts are ne v e r so giv e n to a single mind, and be cause , in conse que nce , it is ne ce ssary th at in th e solution of th e proble m know le dge sh ould be use d th at is dispe rse d among many pe ople . T h e proble m is th us in no w ay solv e d if w e can sh ow th at all th e facts, if th e y w e re know n to a single mind (as w e h ypoth e tically assume th e m to be giv e n to th e obse rv ing e conomist), w ould unique ly de te rmine th e solution; inste ad w e must sh ow h ow a solution is pro- duce d by th e inte ractions of pe ople e ach of w h om posse sse s only partial know le dge . T o assume all th e know le dge to be giv e n to a single mind in th e same manne r in w h ich w e assume it to be giv e n to us as th e e xplaining e conomists is to assume th e proble m aw ay and to disre gard e v e ryth ing th at is important and significant in th e re al w orld. T h at an e conomist of Profe ssor Sch umpe te r's standing sh ould th us h av e falle n into a trap w h ich th e ambiguity of th e te rm "datum" se ts to th e unw ary can h ardly be e xplaine d as a simple e rror. It sugge sts rath e r th an th e re is some th ing fundame ntally w rong w ith an approach w h ich h abitually disre gards an e sse ntial part of th e ph e nome na w ith w h ich w e h av e to de al: th e unav oidable impe rfe ction of man's know l- e dge and th e conse que nt ne e d for a proce ss by w h ich know le dge is constantly communicate d and acquire d. Any approach , such as th at of much of math e matical e conomics w ith its simultane ous e quations, w h ich in e ffe ct starts from th e assumption th at pe ople 's know le dge corre sponds w ith th e obje ctiv e facts of th e situation, syste matically le av e s out w h at is our main task to e xplain. I am far from de nying th at in our syste m e quilibrium analysis h as a use ful function to pe r- form. But w h e n it come s to th e point w h e re it misle ads some of our le ading th inke rs into be lie v ing th at th e situation w h ich it de scribe s h as dire ct re le v ance to th e solution of practical proble ms, it is time th at w e re me mbe r th at it doe s not de al w ith th e social proce ss at all and th at it is no more th an a use ful pre liminary to th e study of th e main proble m.