NOTES OW LECTURES IM PRICE THEORY me The aw of Varisble Proportions We have just gone through in a form vay the various types of supply con Aitions that my obtain. We have coon that the cupply conditions depend on the cost curves of the individual firm. We now turn to the firm to examine the conditions unferlying its cost curves. Our interest here 1s, of course, not in the firm per se but rather ine fuller understanding of the factors determining the supply conditions in an industry. We might add in passing that @ eupply curve 1s a ueaningfl concept only for a competitive industry. Othervise, price alone does not describe completely ‘the conditions of desand facing the individual firm. We mist alto renember that in going from cost curves to supply curves ye have to be on the Lookout for the possible existence of external economies or diseconomies -- economies or @ieeconomies external to the firm but internal to the industry and hence affecting the supply curve of that sndustoy. We may regard the firm as an intermediary between factor marketa vherein it ‘buys resources and product markets vherein St sells products. For the firm the densnd conditions for the product it produces are sumarized in the demand (or aver~ ‘age revenue) curve for ite product. The eupply conditions on factor markets are sumarized in the supply curves of factors of production to the firs. The tectmo- logical conditions governing the firm are sumarized in the production fmetion which shovs the (nuclmm) quantity of product 1t can produce for given quantities of each of the various factors of production it uses. one of the properties ssigned to thie production ftmetion ts generally de scribed as the lav of Diminishing Roturne. This terminology is closely comected with ‘the explanation of the so-called "lav" in tera of fixed and variable factors of pro- duction, At bottom, hovever, the isme in question has 2ittle or no relation to ‘tte distinction between fixed and variable factors; 1t ie rather concerned with Ee D. H. HILL LIBRARY North’ Carolina State Collocetno effect of varying the proportions in which a{fferent factors are employed; and al factors enter in completely symmetrical fashion. Accordingly, {t will perhaps avoid meuderstanding to call it the Jay of Yarisble Proportions. A rypothetical protustion Auction designed Yo tllustrate this lay of variable proportions is given in tebvlar form in Teble 1 anf in graphic form in figure 1. For this example, Let us suppose that only two fectors of production, say A end 3, fare used to produce the product, Colum (1) givee selected values of the muber of units of B par unit of As S.e., of the retio in which the factors are supposed com bined. Let us skip c jum (2) Zor the monent. Colum (3) shows the mumber of units of output per unit of A for each retio of Bto A. Yor example, it says that 1 oneveixteenth as many unite of B of Aare used, then one unit of profuct will be produced for each unit of A employed; if equal musher of units of B and Ay then 25 units of produst will be produced for each unit of A employed. ow the mere possibility of making statements of this Kind already oays = ‘great deel about the character of the production function. or 1¢ mignt be, sy ‘that one unit of Band one unit of A would produce 25 unite of product, but ovo unite of B end ty unite of A vould produse efther more or Je ‘than 50 unite. In ‘that case the Imovledge that equal numbers of units of A and B vere employed would not be cnough to determine the output per unit of A; in adéition, one would have to imoy the ebsolute mumber of unite, Output per unit of A will be a function solely of the ratio of the factors of production 1¢ and only if the production function has ‘the property thet multiplying the quantities of all factors ty a constant will miltiply output by the sane constant -- e.g+, doubling quantities of all factors Vill double the output. Pimetions having thie property ere by definition hono~ geneous fimctions of the first degree. So our {Lustrative table is dram for such a function. We shali @iscuse the meaning and significance of this property later. Yor ‘the mosent, it will ouffice to say that ve want ultinntely to distinguish detveen two sete of considerations affecting the costs of an individual firm ‘the peoeple @)| @o] © 28 ope 6 (8) 2A a] @) G 3 20) o | ° = ind 3/a6 | 36 L 3 we | & 2 =| us we. ® aa |e é Fame 6 2 =2 ° iz |e 36 2 | ae a [=a af 2 2 25, 2 uf a L aja] sat 2 ae 6 28 el 2 of] =y z x afk 6 2 sak =5_|-xja| sto 8 18 x me | 5 =3_| ans |e a6 | a/s6[ 36 2 ea ma. | a | ane] as = [es Ta. ° Yoter Ini. stands for indeterminate. 2) We. of units of B per unit of A (2) No. of uitts of A per unit of B i 5) ‘Verbal descriptions of colum heaténge; (6) Change in no. of unite of B per Product per uit of A Profust per unit of B Ghange tm product per unit of A unit of A Marginal prod. of B % 9 unit of B (20) Marginal product of A Change in product per untt of B Change in no. of unite of A per— ox seuoyaapa0sd STquTsMA Fo AUT ouy Jo oTdMEY FT am 4 2 g 4 £ a 6 ava B Z G 3 q ; : 3 toet Ween 6 hp ane rte 1 2f need tt | sc 2 porooxs porte «30 shun ss arses] ——| 39 sun ea aps — - | a 7 | n ~4. a / ~ L gevlso arma dpa somo , / el -_ “ato e 7 \7 v fo asmpord rina) | —- po - a o | _do-- a0 aproome well _b-- | i : a vq og eummgac pouSans oxrya80Kt cued oesaxe Supeveroeq TRUS aayaeog “CPD “s sq woe ¥ 04 sumyar sumgex siaseay Supsvosou YO} Emo “Ay “Od a8 4e portions tn veh 14 combines ftctore; the eonle on ich tt perstes, the La of vaciaste proportions dane vith the ttost et, ant ve can beet abstract fran the toe fumes of scale by rovisoaaliy mgposig At to ave no tafiuences yhteh £0 pee cisely what tnvolvet in epoting the fim production function toe hoogene- oe of the fine Gegres in A aol 3; tot A ed 3 20 be thn cly te factors of roe Auction ‘volved. We abald see, further, thet the influmce of scate can iteit te ‘ew aa the omlt of the operation of thn inv of varlable pepertion, 40 ve are nating 0 1e00 pect aanaption than might ab first be exgnoed. ven tint the rodsctlon Action is hmcgeneae cf the fire dagree an that onty tv factors are involved, pats of solums lke (1) anf (3) describes 18 completely Sf the ontrfon are oufeictentiy munerous, Yor constder the general qe tones ov mich x can be poficed if tare are a; unto of A and by tmite of B. ie ansver can be obtained by commuting Ht, entering it in column (1), finding the corresponding eatry in eolimn (5), and multiplying the remit by a. Thie ta vat ye mean ty anying that in thts case everything depends only on the proportion tn viteh the different factors axe cxbined. Tt follove that atl the rest of Mable 1 can be obtained from colums (1) and (3); end eaumination of the column headings will confirm this: colum (2) te simply the reciprocal of (1); colum (4), equal ‘to column (3) divided by colum (1) or milttplied by colum (2); and so on. ove eaten for putting Sn both estima (1) and (2) ta to enmhie us £2 tras tate ade tae roedty Sato tem of variable end final factre, agpose the fire ‘mist use one unit of A, but cen use varying amounts of B. Then colum (3) == or protuct per watt of A=+ in otal" products colum (8) = or protuct per wat of Bow i “average product” of the "varteble” factor; and calum (7) -- marginal Product of B—~ is "marginal product” of the "variable" factor. Similarly, if the ‘firm must use one unit of B but can use varying mounts of A, we can take colsmm (2) to show the anount of A used. We shall then, of course, vant to read the ‘able fran the bottom up, since this will correspond to increasing axomta of the “variable” factor, Colum (1) = or product per unit of B-- is thn the "total“ke product", colum (3) =» product per unit of A+ the average product: of the “var lable factor"; and colum (10) -» marginel product of A-- the “aarginal” product, of the variable factor. Tet us now turn to the munerieal values in the table and the graph. ‘The particuler eximple 1s set up oo as to illustrate most of the cases thst are arithme- tieally possible within the freneyork of @ tyo-varlable homogeneous production func- ‘tion of the first degree. Yot ell cases are of course erithetically possible -- tor emmple, average profuet cannot inorea fas the relevant variable ineresses and at ‘the sane tine be greater thin the corresponding marginal product. Tn checking for ‘this kind of internal consistency in the graph, {t ehould be kept in mind that A deoresses relative to Bas one goes fran left to right, and that hence in inter= reting the A curves they should be read "backyards," as it vere. (The terme "increasing returns" and "Qiminishing returns" are sometimes ue0d to refer to marginal returns, sometines to average returns, so It vill be best to indicate explicitly which de intended. Furthermore they alwys refer to the be havior of returns as the quantity of the corresponding factor incresses. Marginal retums to B increase at first; thereafter Aiminieh; and ultimately becone negative. Average returns to B increase over @ longer range ~» until @ ratio of 1/kof @ unit of B per unit of A, if ve stick only to the designated points and avoid inter- polation, are the sae at a ratio of Bto Acof 1/2 ae at 1/k, and then diminish. ‘A behaves of course in the samo vay, as ve shall see most readily if we read from ‘the bottom of the table up, or from the right of the graph to the left. Marginal, returns to A increase to somevbere between 1/16 and 1/8 of a unit of A per untt of 3B, then decline and ultimately become negative. Average returns increase to 1/h of ‘a unit of A per unit of Bs are the same at 1/2 a5 at 1/k, end then diminish. Whe first and last entries in the teble deserve a vord of explanation. The product per unit of Ais set at O for B/AE-0; this imlies that B is an “essential” factor in the sense that no output is possible without some B. Since colum (4) ts colum (3) divided by colum (1), the corresponding product per unit of 3 48 0/0, hence indeterminate.= Tt could be that some product could be produced by uee of A clone. In this case, the first entry in colum (3) vould be finite, and in colum (k) So . ‘Sintlar renarks apply to the last entry. ‘the table and graph supposedly sumartze the tecmnological conditions govern tng the production of the produrt im question. ‘That 1s, they are designed to anmver the technological questions: given specified enounte of.the tuo factors of pro @uction, vist is the macimm asount of product that onn be produced. Let us now soe hoy ve would use thie information; in the process, we can also test vhether ell. ‘the arttimetically possible cases they contain are economically or technologteally relevant. Buppose, for example, that we have eight untts of A and 6h units of B. The table shows an output of 32 per untt of A vhen the ratio of D to A is 8 to 1, which vould nea a total output of 256. But is this really the best ve can do? iAurther examination of the table suggests that St 4s not. If tt costs nothing to "throw" B avay -- not to "use" it —= we can get an output of 36 per untt of A, of 288 in all simply by using only 16 or 32 of our units of 8, that 1s, either 2 or 4, units of Boper unit of A. If the table hnd more entries perhaps sone mmber between 2 and 4 ‘would be even better. Obviously, the s{tuntion 4s the eae for eny larger mumbér of unite of B per unlt of A-- so, no matter how plentiful B is, £t will not de sensibie to use more the h unite of Byer unit of A. imilarly, suppose we had the seme 8 units of A but only one wait of B. The entry under a ratio of B to A of 1/8 shows an output of 4 per unit of Aor 32 in all. But again this 4s not really ‘the best we can do -—' suppose we vere to "throw" sway, L.e-, not use, of the unite of A. We should then be opersting with e ratio of B to A of l/l, for which the out put por ult of Ate 9, or mulbiplied by the k units of A being used, total output 4s 36. In consequence, no matter how "scare able to use less Bis, it 4e not eer ‘than 1/ of a unit of B per untt of A-- or stated in reverse, no matter bow plentiful A 4s, it is not senstt le to wee nore then b unite of A per unit of 3. Suppose nov that the ratio of B to A 46 between 1/h and 4, cay B unite of A end 86 unite of B, or e ratio of 1, doce anything sizllar occur? Clearly it does not. BY using all of the A and all of the B, output per untt of A ta 25, total output de 200. ny using lese of the A, say ony b unite, output yer unit of A can be raised to 36, ‘but atnce only buntte are used, total output fe reduced to Uhh; einilarly, by using ess of the B, say only unite, output per unit of B cen be ralset to 36, but only at the expense of reducing total output to 1M. ‘hese examples show that the tree regions marked off in Figure 2 sccording to the tebavior of average rete have very different meening and significance. tn ( ‘the st region, average returns to B are increasing, average returns to A, Amin Lehing; in the second region, average returns to both A and 3 are aiminishing. ‘The third region te the comterpart of the first -- average returns are tncressing to one factor, in thie case A, and Giminishing to the other Now our examples show ‘that the fret and third regions are ones to be sumed. Fut differently, the lures entered in our table for these regions, while aritimetioally posathe, ave tectmologieally Anconaiatent under our assumptions vith those entered eleevhere. The table purports to shov the maxtmn cutput technologically possible for different -coubinations of factors. But it doas not do 40, for, es ve have seen, ven the ratio of B to Ade 6 to 1 there ie a vay of using the factors that will produce an output of 36 per unit of A and hence of b 1/2 per unit of B, vhereas the table shove an out put of only 32 end respectively. In other vords, on technological, grounds alone, ‘the table is wrong, given the sesumptions thet the production function is honogene- ous of the first degree, and that A and Bare perfectly aivisible (this point is discussed below). yor B/A © 1/16, the entry in colum (3) should be af, in column (), 365 18, for B/A= 8, the entry in colum (3) should be 36, in colum (4), i tor B/A entry in colum (3) should te MZ, in colum (1), 36 tor B/A = 16, the entey in colum (3) should ve 36, in colum (4), 2b. ‘hts then is the Inv of variable proportions relevant for eccncnies: insofarfas possible, production will take place by the use of such @ eoubination of factors ‘that the average retums to each separately vill dimintsh (or at most renatn constant) with an increase in the anount of that factor used relative to the anaunts of other factors. ind this "lav" de not @ fact of mature, in the sense thet nothing else is possible, or that it is desonstreted by repeated physical experinentes it 16 a maxim of rational conduct. somevnet paradoxical that "increasing returns," vich sound 1ike Te may se soucthing good, should be something to be avoided, ‘Tats appearance of paradox may be reduced ty noting that im both the teble end the graph the region of increasing average returns to one factor coincides with negative marginel retums to the other factor. ‘This {8 no accident; 1% ts a necessary consequence of the fect that the prov Auction function 49 homogeneous of the first degree, ao can readily be denonstrated. Suppose that 1 untt of A plus B, unite of B produce X, units of product and that this 4s a region of increasing average revumns to A. Then 2 unite of A plus B, unite of B vill profuce nore than 2X, units of product, say OX) X-vhere x >0. But Decmuse oF hmogeneity of first degree, 2 unite of A plus 28, unite of B will produce only 2x, unite of product. Hence the additional unite of B have diminished output, so B mist have negative margins! product. Tae common saying == there's no use going farther because you've already reached the point of diminishing returns -- is highly misleading. ‘The point not to be exceeded is the point of vanishing (marginal) returns; the prudent man vill seek to exceed the point of diminishing (average) returne.* stove that the equivalence between increasing average returns to one factor and negative marginel return to the other is valid only for a hosogenecus function of the first degree. Suppose the production function 1s homogeneous but not of the first degree and contains only two variables. If the degree of the function is le: ‘than one, then dncreasing returns to one factor implies negative marginal returne 20 ‘the other but the converse does not bold: negative marginal returns to one factor 18 consistent vith diminishing average returns to the other. If the degree of the Tunetion ts greater than one, negative marginal returns to one factor implies in creasing average returns to the other, but the converse does not hold: ineressing6 average returns to one factor de consistent with positive marginal returns to the others Can entries 2ike those in the first and third regions of Teble 2 and Figure 2 ever be relevant? ‘There are two sets of circumstances under which they can be. The first 1s trivial and involves only a verbal exception == suppose that “using” & ‘factor 16 paid for, i.e, involves a negative cost as, for example, vhen it involves ‘using laborers vho are learning a trade and ere willing to pay for it, It may then tbe worth going into the region of increasing return to the other factor and negntive return to this one, Dut in that case, the firm is really profuctng tvo products ‘the output entered in the table, and education, and the table 1s not a complete sum mary of profuction conditions. Another exmple of the sae case is vhere it coste sousthing to “throw avay" a factor, but again this mst mean there are either other factors of production or other products involved. ‘ma more important case ds suggested ty the qualification insofar se posable 4in the ctatement above of the lav of variable proportions. Tt may not be possible for a firm to get into the region of @iminishing returns for either of two reasons: ‘yecmuse the quantities of relevant factors of production are outelde of its control, for because of indivi sit Jet us postpone the first reason for the tine being and consider only the second. Suppose factor Ais lend, plus labor, ete., in fixed ratios to the ancmt of land; factor B, services of « tractor in cultivating 1t5 and ‘the product 1s, say, vheat. Suppose, further, that tractors cone in two sizes, one of which, size Tt, can be regarded as “brice” as much tractor as the other, or size I. Yor a given anount of Factor A it may vell be that total output ie less with one ‘actor of size If then with one tractor of size I, say becmuse the eualier tractor does enough vork per uitt of tine to cultivate the given area with the given other factors, so the only additional effect of the bigger tractor 1s to trample dow more of the thet. "ais nonns that with the Bigger tractor, we ere in the region of neg~ ative marginal returns to tractors, inereasing average returns to land. Yet if onlyos the bigger tractor 1s available it my be better to use it than no tractor at all. n thie ease tt is not physically possible to throv “half” the tracter avay, though At yguld be desirable to do so. Yote that this effect does not.come from ovming, the ‘tractor rather than renting 1%; Shp sane effect arises if tractor can be.rented. by ‘the hour, say, but the only tractor that can be is one of size II. For using this: tractor ml? the time may not be equivalent to using a tractor of size T all the ‘Vine. he number of "tractor days" of eervice that can be used may be perfectly continous, yet indivieibtiity may be present, Note also that indivisibility of one Taotor means increasing average returns to the other factor, not to it. In the particular example, the dndivistbtlity could presumably be removed on ‘the narket by eelling the larger tractor and buying a emaller one- But it 4e clear that this may not be possible, since there vill be sone minimm size or scale of tractor made. Ultimately, met such indivistbility treces to the Infivistbility of ‘he maman agent (the absence of the "Balf-eize man" to drive or sake the "half-eize tractor" Translation of law of Varisble Proportions into Cost Curves Tet us now turn to the determination of cost curves from a production fimction Like that, sumarized in Teble 1. Suppose first thet there are no iniivistbt}ities and toqt the firm ie perfectly free to hire any number of units of either af the sctgn of reduction. ‘here te nov no definite maber of nite of each Zeotr of totustion avalable. Instat, the fire 1s Lntted ty the price cr tmler ance sony, tin my cure) of the factors et potunton. Aemme competition inthe factor mrint, and expe the price of B ie zero. The te analogous tom ube sinited smnunt of Being sratibie ea obviously the opis exthnniion of A Vi be between 2 ant untes of B per nit of A, ls wil mean en oxat of 36 per walt of A or a cort of a per ait of yrotuty vere Py is the yroe yer wait of A. Cleary, wer the given aumptions, ‘Ade costs independent of tpt, so fhe cout curves will be bortsontal, as follow-10- Cost per uit ee ———————— Quantity per unit of tine Simtlsrly if P, vere sero, but ,, (the price per unit of B) yore not, the cost vould be 2 anf 2 toh unite of A would be used per unit of B, Suppose nov that neither prize 1s zero. We iow from cur earlier analysis that the optimm coubinetion vill ve cea en ep BR Yor example, suppose P, * $1.40, P, # $110, then the optimm coubination would be ‘between 1 and 2 units of B per unit of A. For 1 untt of A to 1 o! 2, the cont per walt of product vould be 10}; for @ unite of Dyer wit of Ay 0fs for 4 alte of Dyer wilt of A, 36 1/94. Again the marginal and average cont curves vould cotnetde an tn the above figure. ‘he axalyeis up till nov bas show that if all factors vere perfectly aivien thle and obtainable ty the fim at a constant supply price, then the optimm combine ation of A/B would be the sane for all levels of outyut. he marginal and average coet curves would then be colnctiest and thetr hefght would be deternined by factor peices. ‘ais case {0 not, Dovever, the only relevent one, or even the most signtfi- cant one. In the first place, horizontal cost curves would imply either monopoly (4t the height of the comt curve were lover for one fir than for others) oF com plete indeterminaney of the size of firms (if several or many firms had curves of ‘the same betght). Tn the second place, St ts not waeful 4a analyzing different "rms", watch are distingutshed precisely by the Aifterent possibilities of changing ‘the exomte of various factors, Want this case does bring out is that, for honogen- cous production functions of the first degree, rising cost curves, hence linttationsfon the eize of firms, must be sought in Miwttations on the firm in the possibility of varying the anounts of some factor or other. Suppose that the supply of A ie fixed to the firm at one unit -- either temp- cverily for © short run probles or permanently. he firm can then vary its output only by varying the snount of B employed. Ite cost conditions cen then be derived Asrectly fron Table 1 together with (1) the price of B, and (2) knovledge vhether ‘the unit of A is divisible or not. ‘he folloving table and graph gtve the results when the price of a untt of B ts $1.20. Woother | de indivietble makes difference only for eal enounte of B, for clearly B is taken to be divisible, so vhen large amounts of B are supposed employed ‘theze ds clearly nothing to prevent some of the B from not being used. For enaller amounts of B, when A 1s indivisible, the figures in the original Teble 1 are rele~ vant; when A ie divisible, the revised figures that take account of the possibility of not using some A, ‘.e., of not letting the ratio of B to A in use fall below $. ‘The marginal coste can be calculated in either of two ways: by dividing the ‘norenent in colum (+) by the corresponding insrenent in colum (2) oF (3)s by eaviaing the price of a unit of B by Ste marginal profict as shom in colum (7) of Mle 1, for A initvistble = or in en appropriately revised column, for A eaviesoie. ‘Woon B/A is between 1 and 2, ve have the combination which tural out to be opti in our earlier example of both factors variable when P, © #40; P * $1410. Since the price of B As asmuned the sate in this example, the marginal cost, for tha: coabination of factors, 16, of course, the sane as previously, or 10f per unit. The dashed Lines in the graph are for A indtvistble, the indivistbtltty pro- duces a decline in both average variable costs end aarginal costs, the counterpart of increasing average returne to B and negative marginal product to A. That the decline in marginal costs, or even ite being lover for a segment than the marginal cost when A is Givistbie, te no advantage ts clear fron the fect that average var Sable cost during this interval 4s higher when A ie inéivietble than vhen 1¢ 12one ebie 2 We~ oF ORE a WaT- |n Co__ovaras var Tab oe umits (“A ] A ]able cost [2h 7 or) of. | snase | atvi-| (2) x§i-20| snatvi~ [atv | anatvs- | atvistoie emloyedvicibie| tbl sible | siete | sitie o Te Tol @ @ |e o @) ° o fo ° Tad 03 1/38 1.08 7/8 f.03 1/38] wise [a [enal 0.06 7/8 06 7/8 | 032/38 02 7/2 .03.1/28| ve | * [aael 0.3 6/6 03 7/15 |_.03 3/18 T -02 aft | -03 3/8] i [3 0.27 8/8 03 2/18 -03 1/18 ve | 38 055 | 93 1/38 [ret i 3 110 “os 8/5 20 2 36 2.20 06 2/9 5 a 36 ao 32 2/9 oo 8 36 8.80 -2k /9 6 36 31.6 28 8/9 - oo | 6 3“ue. [eave foxe / Ly bee nef a orvdsrone ho 2 30 35 20 6 10 Price Hi T 1 3 TREE wo a orvfsreabyLetbie. For A divisible, the marginal cost and average variable cost are horizontal, (end therefore coincide) initially. his ie because the Himttation on Ais tor thie interval irrelevant; thie is eeeentially our earlier case when A was & free good becouse dn this interval st ie not worth whtle exploying all of A. To put thie in other terns, the eupply curve for A is talon to be as follows Price of A |. «Supply curve of A aantity of A per unit tine Yor low outyute, the horizontal seguent of the supply curve of A is relevant. Homogeneous #4 sedegres Production Function -- Problem of Sosle ‘me exmmples just Gecussed indicate that the use of a production function that 19 homogeneous of the first dagree 19 compatible with almost any Kind of cost conditions with declining average variable costs if there ere indtvistbil- ities, with rising average variable costs £f there are linttations on the quantity of one factor exployed. Indeed, it begins to look as if a homogencous production fumotion of the first degree can be vieved, not as an onpirically special kind of function but az a manner of speaking shout all functions, as a framyork of refer ence, or tautology. ‘Mle te one vay of vieving 4%; and an extrenely useful vay of doing 90. ron thie point of view, the concept of @ honogensous function of the first degree can be considered equivalent, on the one hand, to the concept of a controlled ex- periment, on the other, to the concept that the units chosen for measuring quantities are irrelevant (the prineiple of relativity). Fundaxente2 to science is the con ception that {f an experiment is repented under ‘dentical conditions, it vill giveay Adentical recults. Bat {6 not doubling the quantity of each of the factars equi alent to repeating an experiment! If the inttial bundle of factors yielded x unite of output, mst not an identical bundle under the same conditions yield X also? Hence, mst not the tyo bundles together yleld ox? Or A¢ the tyo bundles together ‘then X, mst not thet yield 2X, vitle tt 16 said that one bundle alone yields Le mean that the conditions vere not the case and the experisent vas not really the suze experiment? If the one-bundle experiment vere a precise replica of the ‘vo- Bundle experiment in all deta: » except uniformly on half the scale, mst it not yield Xt or to turn to the other argument =» from dimensions =» can anything be considered shanged if ye look at objects through telencopes or microncopes? If ve change units trom rates of flov per veek to rates of floys per montht If ve think of honogeneity of the first degree az a truien, it cannot of course be contradicted. Yet certain obvious examples seen to contradict tt, such as the parable of the fly, which, {t is said, if 1¢ vere reproiuced accurately on 8 larger seule vould be unable to support its om weight. "he eneyer is, of course, ‘hat there must be come "relevant" factor of production that hes not been increased 4m scale elong with the fly's dimensiones in thie cage, presumnbly the alr pressure and the force of gravity. Im the same votn 1s Pareto's anever to sousone vho said ‘that doubling the subway eysten of Paris vould not yield tvice the return (or per haps involve twice the cost) -- for homogeneity of the first degree to be relevant, hho said, there yould have to be two Paris's. ‘The usefulness of this tautology depends of course on the value of ‘the classification it cuggeste of the things that may effect cost conditions. Tt leads to a classification into (1) those that operate through explicit changes in the proportions aiong the factors of production, the chief of which are of course prices (or conditions of supply) of factors of production; (2) those thet operate through Limiting the quantities of soue factors of production available to the tim -- these ‘account for rising cost curves and inclute the existence of conditions effecting cost (size of cities, mount of cosl in ground, constant of gravity, ete.) outside“we the control of the {Individual firm, linttetions imposed by contrectuel arrangements, ‘and. those largely anoaysous conditions concealed in the notion of "entrepreneurial capacity"; (3) those thet produce inttvistbiitties ‘toese account for the pose sibility of decreasing cost curves and in most cases can be ultinately traced back to the indiviatbility of the himan agent, as is suggested ty the fact that the gains from division of labor and spectelization of function are ali included unter ‘hts heading. of course, conceiving of the unterlying production function as homogeneous of the first degree does not imply that the production function ae viewed by the firm 1s homogeneous of the first degree, The firm te only concerned vith thoi factors of production, or other conditions effecting costs, over which 1t as con ‘rol. Tn consequence the produetion function to the fim can be regarded as crove-section of the ulerlying production fmction -- that 1s, as obtained from ‘the underiying production fmetion by giving to the variables over which 1t hae ‘no control the constant values hich they have for the problem in question. Indeed At {6 precisely this step that enables us to conceive of rising longarun cost curves for individual firms and hence to rationalize the extetence of linits to the size of firas. This is vhat vas meant earlier by the renark thet the "scale" of firms can itsele he regarded a rationalized by the lew of variable proportions. Statieticnl Cot Curve Studies ant Output Flexibility A considerable mmber of cupirical studies of cost curves of individual fims have been made within the past two decades. These have been mainly concerned vith estimating short-run curves. Most of them suggest that short-run marginal cost curves are bortzontel over the usual range of output, vhereas the preceding analysis would rathar cuggest rising marginal cost because of the existence of Limite to the amounts of some factors of production even in the long run and certainly in the short run. In an exellent discussion of these stuiles snd sone of their impiic! ‘lone, Tae Apel* points out thet the statistical evidence for thie conclusiona5. * "Marginal Cost Controversy and Tts Implications," suerican Hsoncate Review, Decen~ ver, 19N8, pp. 870-85. te quite Lintted and not particularly representative. Tn particular, mich of the evidence is for pertods in vitch output vas relatively lov so there might have been Numsed capacity"; L.e., in temas of our preseding analysis, periods in which 1 was possible to keep the ratio of factors fixed when output wae increased dempite the Limited quantities of some factors, because it had previously been rational not to use part of the latter factars. But it 4s not at all clear that the results can be entirely explained in ‘this vay. Tn any event, consideration of these statistical results led George Stigler* to suggest @ force, hitherto neglected, that might make horizontal short * "Production and Distribution in the Short Rin", Journal of Political Boonowy (suns, 1939) gp. 312-22). ‘run marginal cost curves « deliberate objective of maximizing behavior. This foree 4s the desire to cbtein flexibélity. When a plant ia Dullt st ie not expected thes gectsely @ single output will be profuced year in and year out. Tt ts Jmova thet ‘there will be fluctuations in demand and in desired output. ‘he problem in other words {5 not to miniaze the cost of « given output steadily and regularly to be produced but: to mintaize the cost of probebility Aletribution of outputs, tm ‘ALeating the fraction of tine each output vill he produced. ‘The relevant varie be to measure along the horizontal axte te not "the" outyut but the average” output taking AU account of variations fren that output. Yor example, consider ‘the average variable cost curvog show in the following graphs. MWathod of prow duction A 1s a highly rigid method which 4s highly efficient for perticuler out~ ‘ut but not for axy other. he A curve shove the average cost if precisely ‘the output indicated on the horizontal axte te produced day after day. The A’ curve shove the average cost if the hortzontal axis 1s regarded as the averageab output orer time and actual output 4 regerded es fluctuating from day to day about ‘thie average in come given fashion. The tyo curves B and 2’ have the corresponding meaninge for a "flexible" wethod of production. Yor the graphs as drava, it is clear thet the better method of production for @ given unchanged output 1s Aj for a distribution of outputs varying from day to day ercund %, B. Pe P At SS a Method. B ajote that A’ need not be shove A everyube a5 in this graph. In general, A’ vill ‘ve above A, the same as A, or Belov Act any point according as A Le concave upward, Mneer, or concave dovmvard, for the relevant region about this point. Derived Denana The Giatinetion generally made between the theory of the pricing of final products and theory of the pricing of factors of production ie something of « carry-over from the early division of economies into tvo parts: Velue and Distribution. ‘The ‘theory of value concerned iteeIf vith the prices of fins) products and the thecry of distribution with the prices of factora of profuction, primartiy as e guide to understanding the division of the totel product anong major social classes (bence ‘the designation —- distribution). he theory of general equilitrun merged theve ‘two inquiries as parts of one pricing problen iovolving the simitencous deternination of both sets of prices. At the same tine Marshall's exphasis on pply end denand as an “engine of analysio” rether thm on the substantive thing enelyzed mde Lt clear that the same analytical apparatus te applicable to thecha pricing of final products and of factors -- in both cases the problex ean be ex pressed in terms of denand and supply end the cructal question te what determines ‘the shapes of these curves. Is {6 here thet the pricing of final products and of factors of production Aifters, he demand for final products reflects directly the "witlity" attached to ‘then; the denand for factors of proiuction does so indirectly, being derived. tron ‘the denaad for the fingl produsts. ‘The Link between the demand for the final produrt ant the demand for factore is closest vhen the encunt of the factor required 1 rigidly and technically Linked to the snout of the produst. So before proceeding to a general aulysis of the denand for productive services, ve will find 2¢ help- ful to consider thie spectel case, which Marshel. deals with umler the heading, ‘the Theory of Joint Demand. ‘The theory of Joint demand begins vith the notion that the demand for the final product 42, in some sense, @ joint denand for all the inputs. ‘This notion Deconen more than a trite trufm if we assume fixed proportions; that the product can be made only by one unique proportion of A/B. Fron a descriptive vleypoint, such a state of affeirs ts hardly typical. However, analytically 4t te a useful shetrectiion for many problems, especially of a short mm character. Keeping in mind this assumption of fixed proportions we shall now proceed with the construction of a derived demand curve. Let us comme that 1 handle + 2 blades = 1 knife.Price per Inife Price per handle Price per two blades ‘blades Ls Branties B vee > Fy o 2 2 3 4 5 6 7 --- Quntity of mives per unit tine 0 1 2 3 b 5 6 7 =~ Quantity of Randies per unit tine © 2 & & 8 20 a2 vb -- Quntity of blades per untt time ‘The diagram gives the denen for Imtves and the supply of blades ani handles seperately. ote thet the scales must be drawn appropriately if the curves are to bbe comparable: for blades and handles, the unit mst be the quantity required for one Yate, Yor this reason the quantity scale show for cach mmber of Imives the aac number of handles but tice that mmber of blades, Statlarly, the price scale stoxo price per knife and per handle but the price per two bindes. With these scales, and given fixed proportions, £t 4s obvious that the supply price of a knife ‘for sxy given quantity of knives 4s equal to the supply price of « handle for the name quantity of handles plus the supply price of tyo blades for twice thet quantity of Dindes; these supply prices are the minima prices et vbich the tanfle and blades required for a Imife vill be fortheoming; hence, 1f ve suppose the cost of ascenbling to be negligible, their sun 1s the minima price at which the corresponding quantity of Imives will be forthcoming. ‘The curve labeled the supply of intves is therefore the vertical am of the two other supply curves, Tta intersection with the demand,as curve for Imtves gives the equilitrium price of ives, and the supply prices for ‘the corresponding quantity of handles and blades give the equilibriw prices of handles and blades. foe can ve construct a demand curve for one of the jointly denanded stens separately? ‘The mximm price per inife that cen be obtained for any given quantity of imives is given by the demand curve for inives. The maximum price per 2 blades for that quantity of blades will clearly be this macimm price for inives minus the ‘minim price per handle thst needs to be paid for the corresponding quantity of handles, and, for fixed ewply conditions of handles, the latter ie given by the ‘supply curve of handles. It follova that the derived denand price per 2 blades 4s given by the vertical difference betveen the denand curve for knives and supply curve of handles, 3 in the following figure. Price per inite Price per handle Price per tyo blades Shenaies _ Metres <— Derived demand for blades Gay per GE ES of 2 3 hk 5 6 7 <8 bros 23 4 5 6 T @ Handles 4 6 8 w woh Dd Bates "he reason ve want euch a curve 46, of course, to trace the influence of changes in supply conditions of blades. The intersection of the supply curve for blades with this derived demand curve for blades will give the equilinetum price of biades, for given suply conditions of handles and denand conditions for knives. In similar fashion, e derived denand curve for handles could be constructed.“0 Wore, hovever, that the two derived deaand curves cannot be regarded as simltansously valid except at the orteinel equiittriue point. For cach assumes the price of the other component to be on its supply curve. A movenent along the derived doumnd curve for hmAles tuplies that the price of blades is being determined by @ nove nent along the supply curve of blades, not along the derived demand curve of blades. Brice per imtte Brice per handle Price per two blades Syreaes Pintves Derived denand for handles Guamtity per walt time oft 2 3 4 5 6 T 8 9 = knives 2 30% 5 6 7 8 9 = Handles 4 8 8 10 12 wh 16 18 % Blades only at the equilitrtum position 1e the demand price for exch component equal to its supply price; hence only at this point are the tyo derived demand curves consistent. The same analysis can be carried through for Joint owply-a oe per appropriate unit keep Peneep | Pstton Proor, Qutity per wilt tine: ‘Sheep mutton Woot Price per uppropriate unit sheep Derived supply of woo! Ayutton ee Quantity per unit tine: ‘Shee watton Wool ‘the supply price of the quantity of wool ylelded by a cheep for any quantity of wool 1s the supply price of a sheep for the corresponding quantity of sheep minus ‘the denend price of the mount of mtton tn e sheep for the corresponding quantity of mitton.nem Mantyulation of these curves yields readily the familiar propositions tbat am increase in the supply (J.e., reduction in the supply price for each quantity) of one of @ pair of jointly donméed toms vill tend to raise the price of the other items and that a increase in the denand for one of a pair of Jointly supplied tens vill tend to reduce the price of the other. ‘Ae in all problens of dexand, the elasticity of the derived demand curve 1s ‘a very smportant property. What factors determine the elasticity of a derived Genand curve? Marshall (Bk. ¥, chap. 6) gives four principles governing the elastietty of the derived desand curve, the derived denand for any factor used in Phxed prom portions with other factors will be more inelastic (1) the more essontiel the factor 4m question -- thie condition is guaranteed in extreme fashion by the assumption of fixed proportions; {ts inelustion is implicitly e generalization to cases in which proportions are not rigidly fixed; (2) the more inelastic 1s the demand curve for tar Fim products (3) ie sm2ter the fraction of total cost that goes to the factor tn question; (Ib) the sore inelastic the supply curve of the other factors. ‘he three final conditions may be demonstrated geoneteically: Condition 2 Price per knife Price per handle Price per tvo blades quantity of matves Blades323. ‘The dashed alternative denand curve for Inives 18 more inelastic at the equllibriua price than the original and it {¢ obvious that so 1s the dashed alter- ative derived demand curve for blades. Condition 3 Price per appropriate unit qunatity per unit tine ‘The dashed alternative supply curves for handles shovs double the origina oupply peice for each quantity. In consequence the denand price for bledes at the former equilibrium 1e less then before. Assume that an appropriately shifted eupply curve for blades left the equilibrium unchanged. Then the price of blades would be a maller fraction of ‘otal price. It 1s obvious thet the dashed alternative derived denand curve 4¢ more inelastic than the original for two reasons: (a) it As steeper co dq/ap ie smaller in absolute value; (b) the price of blades is less so p/a, ty vhich dg/ap 1s multiplied to get elasticity, is smaller.oe onastton Price per appropriate unit Syenates ot @ 3 k 5 6 7 8 ~iepropiste quantity per unit time ‘the dashed alternative supply curve of handles is more inelastic than the original and so 49 the dashed alternative derived denan curve for blades. This analysie will be ost useful In those cases in vhich changes in prow portions of factars are of least importance for the problem st band, ‘This 49 par tioularly Likely to be the case dn problens dnvolving short run edjustment. ‘The Longer the time allowed for adjustment the greater ie likely to be the error inw volved in neglecting changes in proportions ‘the usefulness of the analysis can be illustrated by applying it to the problem of interpreting the effectiveness of unions in altering vages and the cir cumstances on which this effectiveness depends. This 1s a good illustration pertly vecause short-run considerations bulk large in union behavior. Aisouseton of untons that folloys 1s quoted from Mtlton Friedman, ‘Significance of Labor Unions for Beonowie Policy," in The inpuct of the Unton, ed. by D. M. Weight (Harcourt, Hrace, & Co., 1951}, yp. ef=2l5.) ‘me pover of unions, as of any other monopoly, is ultimately Limited by tthe elasticity of the demand curve for the uonopolized services. Unions have ele- nifieant potential pover only if this demand curve is fairly inelastic et vhatven then, of course, they mist also be able to controt etther the supply of workers or the wage rate explayers vill offer workers. Demand for labor ‘the theory of joint demand developed by Marehsli ie in some ways the sost tusoftl tool of orthodox economic thecry for understanding the etreustances unter ‘Which the demand curve will Re inelastic. Te vill be recalled that Marshall en: ‘Phasized that the denand for one of @ mmber of jointly demanded itens i¢ the more Snelastie, (2) the more essential the given ‘ten 1s in the production of the taal product, (2) the more inelastic the demand for the final product, (3) the smaller ‘the fraction of total cost accounted for Wy the Stem in question, and (H) the more Anelastic the ewpply of co-operating factors. The most significant of these items ‘hLered Marshall, Principles of Bronomics (Bth 04; Macmillan, 1920), pp. 385-386. for the analysis of unions ere the essentiality of the factor and the percentage of total costs accounted for by the factor. Nou, a factor is kely to be fer more ‘essential in the short run then Sa the long run. et e unton be organized and let St suddenly raise the vage rate. Bmloyment of the type of lsbor in question is “Likely to shrink far less at first than 1¢ vill over the longer run, when it 1s possible to make fuller adjustment to the change in vege rete, This adjustment will take the form of eubstitution of other Sactore for Ukis one, both atrectly 4n the production of each product, and indirectly in consumption as the increased price of the products of unionized Ishor leads consuners to resort to alternative ems of satisfying their vents. Tats simple potnt 1s, at one and the sine tine, important {a understanding hoy unions can have substantiel pover and hoy thelr pover Se ehseply Mntted in the course of tine. ‘the importance of the percentage of total cost accounted for by the factorLeads one to predict that a union may be expected to be strongest snd most potent hen £t 1s composed of @ class of vorkers whose wages make up only a small pert of the tatel cost of the product they yroduce -- condition satisfied, along with essentiality, by highly skilled varkers. Tae is the resson vhy economic theorists have aiways been inclined to yrediet that craft untone vould tand to be the most potent, ‘This implication of the joint-deama analysis cvens to have been confined ‘by experience. Walle dndustrial unions have by no seans heen impotent, craft unions have in general been in a stronger econeaie position and have maintained it for Longer periods mple though they are, these implications of the joint-Gemant anslysis have considerable value in interpreting experience, primarily because other economic changes frequently conceal from “casual” cbservation the action of the forces isolated in the theoretical analysis. hie point can be exeuplified by « brie? examination of three major apparent exceptions to the generalization that industriel unions sre Likely to be less potent than craft none. In each case, it vill be found that other economic changes tended to make the strength of the unions appear greater ‘than 1¢ actually was. (2) The Untted Mine Workers’ Union sppesred highly suseessful from shortly ‘vetare 1900 to about 1920. This period coincided with a long upead movement da general prices end wages, so st least part, and perhaps most, he eqparent success of the union can be attributed to ts receiving credit for wage increases thut would ave occurred anyway. Seanty evidence megests that wages in soft cosl may have rieen sonevint ore then wages in general during this period, so that all of the age rise may not be attributable to general inflation, ‘The difference may be evidence that the unton had some effect ca wage rates, or may reflect the operation of sti1l ottier forces affecting the swply of ant denand for Inbar in coal mixing, such aa chiges 1n levels of education, in the couposttion of the stream of Amigrants, ete, Tt would toke a for more detatled exmination of the evidence‘than ve can afford here even to form an intelligent julgnent about the relative importance of the various forces. From 1960 to 1933, the general price level vas stable or falling, coal vas increasingly being replaced ty oil, and the United Mine Workers’ Union practically vent to pieces. Tt vas unable to prevent the underlying economic forces from work= ing thenselves out. Yet st least events of the earlier part of thie period are @ tritute to the short-run strength of the union: the union vas clearly responsible for keeping con wage rates from declining for sone time in the face of the sharp Arop in weges and prices generally after 1920. This illustrates the implication of the joint-denand analysis thet the strategie position of unions will be stronger in the short than in the long run. It also {llustrates a not atypical train of events. Attendant favorable eircunstances enable « union to gain strength in the munber and adhesion of its members ty appearing to accomplish more than its basic economic power would permit; the attendant favorable circumstances without which ‘the union might never have survived Gissppear, but the historicel process 1s not completely reversible: the union for a tine et least rensins strong aud capable of preventing the readjustnent that would otherwice take place, though eooner or Inter 1% 16 Likely to veaken and die if other favorable circumstances do not come along. Mais train of events may be repeating itself in coal. Since 1933, prices ‘and vages in general have again been rising fairly steadily, at e particularly vapld pace, of course, during and after the var, and the union has re-established, Ateele. Once again, the union seens to be shoving reel strength less in the wage rises St has attsined than in its prevention of a subsequent rendjustaent. (2) the garment vorkers’ unions -~ the International Iailes Gersent Workers! Union and the Analgansted Clothing Yorkers -~ achieved their initial successes tn ‘the decade prior to 1920, reaching a peak along vith the postwar inflation in 1920. Again, the untons may have made the wage rise somewhat greater than it vould havea een otherise, but clearly a large and probably the major part of the wage rise for which the unions received crelit would have coms anyray. ‘Though these unions declined in mesbersbip and Japortance during the 1920's anf early 1930's, they fared better thea the United Mine Workers! Union, in uy view largely or wholly be- couse of on attendant favorable cireuistance. These untons vere in an industry that had been largely supplied by imigrante from astern and Gouthern Europe. ‘Union or no union, the stringent restrictions on immigration ixposed after the First World War were bound to reduce the supply of workers and thus to strengthen their economic position. ‘he next spurt in union strength came during the period of generally rising prices end veges folloving 1933. ‘this these unions too have Flourished only vhen unterlying econoaic conditions vere generally inflationary. ‘the auto and steel untons (3) the more recent large industrial union in particular -+ have been operating throughout theiy lives dn a generally tn- Fintionary enviroment. ‘The strength that this hes permitted them to gain will wwe shall argue later that they ‘be denonstreted in a somevhst paratoxicel vay were responsible for preventing the wages of their members from rising after the Becont World Wer a5 mich as they yould have in the absense of the union. T doubt ‘that these unions had mich effect on wages prior to 1915. The recent, mch-pub- Adeized agreement between the United hutombtle Workers and the General Motors Corporation eens to me almost « public announcement of union wealness.# the agreesant calls for a stesdy annual increase in the basic rate, plus cost-of-living adjustments, Tn considerable part, these changes are costless to ‘the company, sines, ae experience in the autonobtie industry before unionization and in other indusuries amply docuente, they are the kind of vage changes that come anyway, though they are perhsps larger In ugnitule. They represent a clear case of @ union seeking to gain credit for vast vould hsppen amyvay. Assuring Steel? euch credit in so puble and Grenstic a fashion may be extrenely clever Union tactics) the need for using such tactics 1 significent evidence of basic weslmess. ‘he length of the ogreenent is of major value to the company, which Ye asmured thereby of uninterrupted control of its affairs. T doubt thet = really strong union vould have granted ouch tern: Jn interesting and instructive example of the tendency, suggested by2 Jotmt-desand analysis, for the strategie position of unions to appear stronger in ‘the short run than in the long run is provided by the medical profession. Tn eco nomic estontials, the medical profession is analogous to a craft uton. Tt com sists of a highly skilled group of workers, closely organized, and in an especially strategie position to keep the supply of workere dam through control over state Licensure and, as a consequence, over admission to medical schools. True, the medical profession differs from the ususl craft union in that the return to the worker (aedical fees) accomts for a considerably larger fraction of the total cost of the final product. Hovevar, even this difference can easily be over~ stated; costs of hospitals, medications, and the like are by no means negligible. Moreover, this difference is typically supposed to be counter-balanced by inela: ‘tietty tn the denand for medical care. There is Little doubt thet the medical profession has exereiset its powers on various ocet sons to Limit entry to the profesefon fatrly drastically: over a to enter fmarican medical schoole has been unable to gain edatesion, and it ie clear that the mimber of persons seeking entry 1s considerably less than it yould be if 1b were not for the Imova difficulty of entzy; further, cerfous impediments have been placed in the path of potential entrants trained outeide the comtry. Yet, restriction of entry bas succeeded in raising average Incomes in medicine only by something Like 15 to 20 per cont. Chiropractors, osteopaths, faith healers, ond the Like have turned cut to be imortent substitutes, and the ine ‘for evidence on the use of restrictive practices and on their effect on income fee Milton Eriedsan and Sion Kumets, Incqua from Indapendant Professional Practice (ational Bureau of Heonomie Recoarch, 1505), pp. 0-20, TIB13{- crease in their mmbers has bean one of the zost important effects of the restriction entry into medicine proper, an impressive example of the possibilities ofa substitution in the long run ‘The stort-run effects of restriction are mare n0- ‘ticeable than the mesns whereby the strength of the union is undermined in the Long rum, viich as noted below, is one of the chief factors that leads to an exggeration of the eftect of unions. Supply of Labor and control over Wage Rates ‘mother Line along which orthedox economic analysis has sous interesting iuplicetions is the role of sovcalled restrictive practices. Tt ip clear that if fe union can reduce the mupply of persons evaileble for jobs, 2% will therety tent to raise the vage rate. Indaed, thie will be the only vay of raising the wage rate if the union cannot exereiee aug direct control over the wage rate iteelt. Yor example, in « field Like medicine, there 4s no significant way of exercising direct control over fees charged, or over annual incomes of physicians. The only effective control is over the mmter of plysicians. In consequence, medicine 1s a clear exsaple of the Kind of situation thet 1s usually enviseged in vhich the wage rate or its equivalent is raised by deliterste control over entry into the occupation, ‘This Line of reasoning has 1ed to the view that, im general, unions may be regarded as exercising control over the vege rate primarily by controlling the eu ply of workers and that, in consequonce, the so-called restrictive practice: ‘nigh union dnitiation fees, digcriminstery provisions for entrance into unions, sentority rule, ete, -- have the economic function of reducing the supply of en= ‘rants so as to raise vege rates. This is sn erroneous conception of the function of these restrictive practices. Thay clearly cannot serve this function without ‘2 closed or preferential shop, vhich alzeady isplies control over euployers de rived trom sources other thn control over entrance into unions, ‘To see the fmction of these practices and the associated closed chop, Tet us muypoce that ‘the wage rate can be fixed shove its competitive level by direct means, for exmmpie, by legal ensotment of a nintmm vage rate. hls vill necessarily mean that fever Jobs will be available thon othervise and fever jobs than persons seeking Jobo.“3 sts excess supply of lator mist be @isposed of eomshoy => the jobe mt be rationed aong the seckors for jot And this ie the inportast economis fimction the 2o- called restrictive practices play. They are a nem of rationing the Linitet rusher of jobs exong eager sqplicente, Since the opportunity to verk at a wage rate above the competitive level ine cousiderible eoooate value, 4% 4e unter standable Vint the restrictive prastioes are imyortant and the source of mich él pate. ‘ie question reaaine hoy the wage rate oon De controLieé directly Wy mesns ther than Legal Gnctaent of u xintmn yage rate, %o do thls, unions mist be able to exercise control over employers = thay mist be able to prevent existing exployers trom unlerentting the union wage rate, ae well as the entry of new exployers who would do eo. Thay must southoy te able io farce ell explayers to offer the union wage rate and no less. the devices vinrehy thie ie done are mmeroue and can tary be nully enumerated here. Hovors:) one feature of the various devices wherehy wage rates are cisestiy erfarced or entzy Snto an occupation Liatted is eesential 255 our puryoees, namely, the extent to which they depend on political ecststance. Perhaps the extreme example is aghin aedfefne, Sn wath practice of the profession 4 restricted to ‘hope Licensed ty the state and Hommare in turn is in geoerel ‘laced in the Inds of the profession itself. state Liceneute applies in sintlar ‘fashion to dentsats, myers, plumbers, Denutiesans, Drbert, noriiedane, and & hott of other occupations too numerous to Ist. Wherever there ie Licensure, 3 fe sinost invardahiy in the hande of the exteting meubers of the ocewatien, vio sinost oe invariebly seek to uae 26 to Limit entry, Of course, in many cases, these techniques are largely ineffective, either because 4% 18 not feestbie to restrict drastically the mater of censes granted, or Decnuse 1t £9 possible to cvndo ‘he Licensure provisions. Rut thay do exenpltfy tov poltteat gover can be wed to control entry ainectiy, ony slightly removed from this kind of Licensure provision anf in aany veye far mare effective 4s local political support through<3 Tullding codes, health regulations, health ordinances, and the 1ike, sll of whtch serve mmercus craft unions as a means of preventing non-union workers from engseing in their fields through eubstitution or elisination of materials or techniques, ext of preventing potential exployers from undercutting the unten vage rate. It ts no accident that strong unions are Zound in raflvays, along with federal regulation. Again, union actions involving actual or yotential physical violence or coercion, ‘such as ass picketing and the Like, could hardly teke place vere it not for the mepoken soguicscence of the muthorities. ‘Thus, whether directly in the form of specitic lave gtving pover to union groupe or indirectly in the form of the ainos ts phere and attitude of Inv enforcenent, direct control over union wage re closely consected to the degree of politicel assistance unions can comand. Tere again, there 1s a very close parallel between Iebor untons on the one hand and industriel monopolies on the other. In both cases, wldeqpread sonopolies fre Likely to be temporary and susceptible of assolution unless they can call to their ald the political pover of the state.336 The Theory of Distribution with Pxed Proportions ‘The Joint denund analysis superficially sens to explain the determine- ‘tion of the price of each of tyo fuctore of production that mst be combined in rigidly fixed proportions to produce @ product. But it does so only by taking fas given the cupply curves of each of the two factors of production, Now these supply curves in turn depend on conditions in markets for other products; they reflect the quantities of factors that would be available for thie use rather ‘then in sone other, and hence depend indirectly on derived denand conditions in other markets. ‘The question arises whether the joint denand analyeie can be generalized fron the partial analysis so far considered to a more general, one. ‘What explains the prices of factors of production if every product satieties the conditions of the joint demand anslysis, 1.e., every product te produced under conditions of fixed proportions ("constant coefficients of production” 12 an other vay of describing thie case)? ‘let us first suppose thet the proportions in which the factora are cam Dined 4s the same in all industries, that, say, ib takes one unit of A plus one unit of B to produce one untt of X, or one ult of ¥, ete. In thie case, any ‘tyo “comodtties" are perfect substitutes in protuction, that 1a, the inditfer fence curve (or transformation curve) shoving the various combinations of, say, X and ¥ that can be produced with any given quantities of A and B vill be a straight line es follows for 100 units of A aad 100 units of B. cumntity of x 1100 OO“ Gaanty oe x Chearly X and Y must sell for the sane price in a free market and similarly for the other comodities no matter vhat quantities of them are pro- duced, ‘The relative demands for then will determine the quantities of them pro-<3 ‘the fact that there are different duced but vill have no effect on thetr pric commodities therefore 1s unimportant on the side of denai for factors of prom duction. Since their relative prices are slveys rigidly fixed, it 1s as if there ere only one comodity, say Z- ‘te staple case iLlustrates an important general point, namely, that substitution in production is an alternative to substitution 4m consumption and vice versa et us nov construct a derived dean curve for factor 3 along the Lines of our Joint demu analysis. To do so we need the denand curve for Z and the supply curve of A. How chal ve drav the denand curve for Z, the single comodsty tn the commmity? Our analysis Se concerned vith relative prices, not absolute prices, aince ve have introduced no "money" into the econowy, eo this question Involves deciding on the “nmeraire” in terms of vhich to express prices. Since our fundmentel problem ie the Givieion of the total output anong the cooperating factors, and since, thasks to fixed relative prices enong final products which Justifics our treating them all ae a einzle product, there is no problem how ‘to measure output, it seems convenient to express the prices of factors of pro~ auction in terns of the final products {.e., to take Zas a mmeraire. Bit ‘then the price of Z in terms of Stself as mueraire te clearly unity ty defini ‘tion, no matter bow mich or Little Z there ts. But thie means that (by defini« ‘tion) the denand curve for % 4e @ horizontal line at a price of untty as in the following Aingren- Vat of the oupply curve of A? here < yrecmably some maxima flo of ‘A tamt can be made avatlable to the produetion of this comodity, say 100 per uit tine, If ve tick rigorously to the assumption thet Z 4s the only final product, there is nothing else that these services can be used for and benee ‘they will be evatleble for this use at any price, L.e., the supply curve of A ill be perfectly inelastic gor any positive price, and perfectly elestic at a price of zero. Tb ie dram as OFG in the folloving figure. (he elasticity of the supply curve of factors to the market as a whole reflects the existence of nonmarket uses of productive services, here ruled out by definition.)35 Price relative to ‘the price of Z fe Supply of A Demand for Z F 00 Quantity of Z per unit time quantity of A per unit time By our preceding analysis the dexand for B is given by the vertical Aifterence between the deasnd curve for Zand the mwpply curve of A, which yields ‘the Following denand curve for 3. Price relative to price of Z Quantity of B per unit time Tote that tis demand curve is nearly identice) with the value of the marginal, product curve for B. Given 100 units of A, the marginal product of B is unity 50 long as the quantity of Bis less tha 100, zero thereafter. To get the equl- LMtrtus price of B, ve need to iow the supply curve of B. As im the case of Ay 4 will be perfectly inelastic at any positive price, so can be deseribed by a single mmber. Suppose the muber of units of B avatlable per uit time te lees ‘then 100, she supply curve of B (6. in the shove diagran) wi2i then intersect, ‘the demand curve for B at P, oF at a price of 1, oo the equilibrium price will-36- ‘ve equal to untty for 3, visieh means of couree zero for A (as cen be shown Atrectly by carrying through the sae analysis for it). If the emply of B ts renter than 100 (S'B in the above @isgran) the supply curve intersects the denna curve at P, implying «price of zero for B and of unity for A. ‘These two cases are relatively simple and straightforward. If one or the other of the factors is so plentiful relative to the other that not ell of 1 can bbe used, then, in the absence of coxbinntion (implicitly rulea out in éreving our supply curves) 4% will be a "free" good, Tut vhat if the quantity of B available te precisely the sane as that of A, 4.e., 100 im the example? The supply and dexsnd curves vill then be as follove: Price relative to Supply of B price of Z ctoarly aay price of B not grester than 1 or less thin zero 1¢ consistent with equilitrim, Given the price of B, say Fp, the price of A vill clesrly be Py t= Bp, sine the taal anmunt to be aivied between 2A end 18 1s 1 undt of 1, the noun they produces ‘mie eolution is uerstandable; we have no vey of determining the separate contrtbution of A and of B to the total product, hence no way on grounds of thelr marginal contribution of determining thelr separate economic value, only a Dundle of an A plus 3 fe an economically meninghit unit. The product of ouch a unit is one, co Py f Pao 1. Any values of P, and Fy such ‘that they ada up to unity vill do. ‘here ere an infinite muber of petrs of values which are compatible with this type of equilibrium. Beoncsic forces aa3 oot we such do not dletate a unique pelr of values for P, and Pj. They mere! Maite, doe, that Py / Py #1. he actual values of Py and Py depend on other factore. If no “nonecononie" considerations are relevant, it is irrelevant how the totel of tunity te divided between a partnership of an A end a3, tor only ‘the conbined uit ¢ significant, Just as it e of Little significance what pert of the wages of a man is to be attributed to bie right had and vhet to his lett. ‘The provles of the dtyioten of the product betwean A and B is significant omy Af there are non-eeonomle considerations that mike the distinetion of en A from a B significant. In this of 1ey these noneconcaie considerations will completely determine the division -- we vill have the relative retums determined by “pure Yargsining," as 1% vere. We have introduced pure bargaining to explain the atviston of the product etween A and B only when their supply curves coincide. But, it may be asked, may ve not also have to introduce it when the cupply curves do not coincide be- cause the smplicit assumption that there {eno coalition among the A's or aon ‘the Bla will be dnvalia? If, say, the qumtity of A available per unit tine te 250, but of B only 100, cannot the omers of A( ell then A's) secure © return above wero ty forming a coalition? Suppose, for a mment, they do, agreeing to Aside equally among thenselves any snout they get, and suppose for the moment ‘that they moceod in getting nine-tenths of the profust for thenselves, so each of the 100 unite of B (who do not we suppose form a coalition) gets 1/10 of a unit of Z, while the coalition of 150 units of A gets 90 unite of Z. Ia this a stable position? Clearly not, so far as economic considerations ee concerned. Bach A separately 1s receiving 6/10 of a unit of Z, each 3, 1/10 of a unit of Z. Clearly there 4¢ an incentive for an A and a3 to get together ‘outside the comlition. o each A separately, 1t appears thet 1f he leaves the coalition while the others stay, he can bribe « B to depart trom the coalition, and still have something more left for hinself, since the total product of the A and B partmership outside the coalition will be a 3/10 of a untt greater than the-e ‘nm of their returns 2o long as the coalttion is unbroken. hte means that the conlition of the A's is unstable, end that economic forces will be perpetually ‘tending to Aarupt Sf even If Lb once be established. ‘We have so far considered e vorld in which proportions of factors of production are not only gixed in each industry but elso the sane in all indus lurtes, et us now suppose that while fixed in each industry separstely they ere not the sane in ell infustries. As the simplest case we nay suppose two sete of industries. Call the (composite) product of one set X, the other Y, and asome that it takes one uit of A plus one unit of B to produce one unit of Ky and one unit of A plus two unite of B to produce one unit of ¥. ‘These pro- auction conditions will yield a production possibility curve tke that in the following Giagran for 109 unite of A end 150 watts of Bs a Bi rcept eb Fy, not all untts of Aor of Bare weed. Bebyeen ¥ and P,, sone untts of Acre menployeds betveen P, and X,, some unite of B. Clearly in either of these sectors ve are back in cur earlier problem. Between Y, end P,, the price of Awill ve zero, the rate of substitution of X for Y will be fixed by the umber of unites of B required and will be 2 unite of X for 1 unit of ¥, so the price of ¥ will be tilee the price of X. Betyoon P, and X,, the price of B will ‘ve zero, the rate of substitution of X for Y will be fixed ty the mumber of units of A requized end vill be 1 unlt of X for L unit of ¥, 60 the price of x will equal the price of Y+ Whether the final equtlibriun will be dn one of these sectors will depend on conditions of demand. If we suppose the above diagram3 to be for one individual (Im a soctety, cay, of identical individuals), ve can ewerinmoee on it the conewption inittference curves of the individual, whieh yields the three possibilities sumartzed tn the folloving @iegrans. | 1" mr Y In (Z) the point of equilibrium involves the mmawployment of some A, hence price of zero for Aj in (II), the uneaployment of sone B, hence @ price of zero for B. hese are essentially our earlier case. In Tit 1s as if ve had fone commodity, the quantity of vhich wae obtained by tresting tyo unite of X fas equal to one unit of Y; im TT, as 1f we bad one comodity, the quantity of which was obtained by treating one unit of Fas equal to one unit of Y. Ih etther of these eases, demmnd ae it vere determines only the relative quantities X end ¥, ptoduction conditions determine relative prices. Me interesting case 4s IIT. Here production conditions determine relative quantities, demand conditions relative zrice. he price of ¥ te eoueghere box tween the price of X and trice the price of X, the exact polat depending on what price ratia will induce the public to consume the same amount of X aa of Y. Suppose that it took a price of Y 1.6 tines the price of X to induce the public to consune-the cane anoint of X as of Ts at yy Hyp Tay By be the price of X, Ys A, and B respectively. Tt would then follov that: WMatm eR 2) ng f 2m” 16D, or, ubtracting (2) from (2) PF, whteh, fron (1) meansoho ‘tose grices are equal to the marginal product of A mn 3 respectively at the margin, If a wit of A ie added it cen be exployed ty producing one fever mite of ¥, vbich vill release one unit of A end tvo units of B, and two additions unite of % vhteh will require the tyo units of A and two unite of B avatiable. the marginal profuct of A te therefore two units of X minus one unit of Y, or in vanue terme, 2p, - Dy cof Y minus one unit of X, or, in value ters, p, =P, * OP,» Mare generally, choy Stuilerly, the marginal product of B is one unit ‘we can derive the marginal product of each factor, and the value of the marginal. product, for Atfferent snount of it, i.e., we can derive margins) productivity curves, vhich in thte ease, will also be denand curves for the factor. Consider, first, the marginal product of A, given thet there are 150 units of B, If ve ‘think of adding unite of A to the 150 units of B, ve have a choice when ve use ‘the first unit of A whether to coubine it vith 2B to produce one unit of ¥ or with IB to produce one unit of X, or to do partly one and partly the other. Since under these conditions, the rate at viteh ¥ can be mubstituted for Xin production 4¢ 1 to 1 (since Bis superabuntant), the price of X and of ¥ vould ave to be the came if both are to be produced. Hy our convention of taking the price of Kas the museraire, the price of both vill be equal to 1 and so vill total incons. Now at them prices and this income, conditions of demand (utsrity ftnetions") will detersine hoy the first unit of A will be divided ‘Ytetween production of X and of ¥. At one extreme, consumers might prefer ony Y, at the other, only X. In either of these extrene cases, the price of only one of the products vill be defined, but even vhen this is the price of ¥, it will de stuplest, and valid, to regard 1 as equal to 1, Move generally, the consusers vill distribute their unit incowe asong both products, so both wil ‘ve produced, In all thes eases, however, the marginal product of A ie unity ‘at the outset, et us contin to afd units of A. For a time, it is clear, everything is the sane 9p vhen the first unit ie applied to the 150 units of B--ae Bie superabundant, so X ani Y are equal in price, the value of the aarginel product of a unit of Ais unity, the physical product being divided betwean X ant Y in proportions dictated by demnd. How many unite of A su be added before a point 1s reached at which B is no Longer euperabundant, hence no longer @ free good? Clearly this depends on contitions of denand. Tf, st a price of unity for both, X 1s in mich greater Gonand then Y, so the bulk of each increment to tatal output Le composed of % ‘then B vill not becone a "Limttattonal” factor until close to 150 units of A Ihave been added to the 150 units of B available. At the other extrene, if at a price of untty for both, Y 1s in mch greater demand thm X, so the bulk of each Increment to total output Le composed of Y, then B vill become a "Linitetional” factor when clightly more than 75 units of A have been added to the 150 unite of B available. ‘To be concrete, let us suppose that demand conditions ere summarized ty G) x Y ‘This "demand curve" implies that the ratio of ¥ to X depends only on the price ratio of the two products anf not on the absolute level of incone.* If ny = Py ‘the ratio of X to Y is 5/8, which means thet in the initial phase as untts of A 3 the wet of utllity Smotions which will yiplé thle demand omve 4a given Hy ve rad), where FT > 0, are added, 5/13 of each untt 1s used to produce 5/13 of @ untt of X; 8/13 of each unit, to profuse 6/13 of a ult of Y. So long as this eontiques, the msount of B required 4 given ty OF bay ea fet a vhere a is the anount of & enployed, b the exount of B required. ‘This can con- ‘than 150, te, untit ‘cime so long as the snount of B required is lewhee 6) = 150, (6) ae 92 6/1, ‘at vhich point 35 5/T units of X ant ST 1/7 unite of ¥ are being produced, once thie potnt has been reached, further unite of A can no longer be cuployed in this fashion, fn extra unit of A can be employed only by prom ducing one unit fever of Y, and using the unit of A end 2 units of B thereby released together with the efditional unit of A to produce tvo unite of X. Tn physical term, then the marginal product of A becomes two unite of X minus one unit of Y. At the prices of X and ¥ prevailing vhen this point {s reached, Aansly py =, 1) the value of the marginal product ts 2, ~ p, or unity as vefore. But as additional unite of A are aided, the prices of Y snd X cannot remain the sane, for the quantity of Y 1s declining relative to the quantity of X, 50 the price of ¥ mist rise relstive to the price of X in order to induce consumers ‘to buy Y and Xin the yroportions in which they are being made aveil- able, which mene that the value of the marginel product of A declines. Addi tion- a1 units of A will be used to produce to alditional units of X and one fever unite of Y co long as the value of thie combination ie positive, t.e., 60 long ‘as the price of one unit of ¥ ds Lees than the price of tyo units of X. When py decoues equal to A, the value of the marginal protuct of 4 ie zero, end additional unite of A vill not be used at all. In ow special case, when a > 92 6/7, the anount of X produced will ‘be equal to (1) x® ofa - 92 6/1) #35 5/T & 2a = 1595 (8) x = ~(a - 92 6/1) # 57 1/7 # 150 - at se am be Cheskad wast Toting that they Fever to {nterval in which all units of Aad of Bare used. ‘be anount of A used te given ty of y © 5 the emoust of D by x 7 2y 2 150, In the case dn question. Solving these bro equations gives (7) and (8) airectiy.abe Inserting into (3) the prices of ¥ will be sp 8 Gaeu my tm 3 eh so the value of the marginal product will be (20) Yelue of marginal protuct * 2p, = a 8 Banrso] 2 [e-9 8 mee vith be equal to sero vin a 22 + 03 2 The resulting value of marginal product curve is given tn the following graph. The value of the marginal proiuct 4s unity vhen the quantity of Ais 92 6/1 or less, declines at on increasing rate fron 92 6/T to 103 11/13 end te Value of the marginal product of A relative to the price of X 92% 300 103i Guat oF A zero thereafter. If the anount of A available 4e 100, as earlier asamed, the price of As +4 as show by the intersection of the supply curve end the value of mar ginal product curve. Tels curve 1s of course only valid if B ie equal to 150. By exactly the same procedure, the value of marginal product of B can be derived, and you vill find it @ useful exercise to go through the arithmetic of deriving 1. ‘The Anfleterminacy that arose when the proportions were both fixed and thea ame in different industries is entirely eliminated by the extstence of two alter native proportions in vhich the factors oan be conbined, as can be seen from the preceding figure. Tf the quantity of A is less than 92 6/7, tts price 48 untty (tue price of Bis zero); Af the quantity of A is greater than 103 11/13, tte price 1s zero (the price of Bis unity); 1f the quantity of A is betreen 92 6/7 nt 203 1/23, se price te given ty SFG iG, , on the ontinase of the eve plotted shove, ‘Mere no Leger remnine sty sompe for a pure baretatne” Cory of wages.os ‘he Theory of Marginal Productivity and the Dexand for Pastors of Profuction ‘the case Just considered of fixed proportions anong the factors of prow @uetion in each industry separately 1s a special case of the general theory of marginal profuctivity. In that special case, on increase in the supply and conee~ quent reduction in price of « particular factor increases the quantity of the factor denanded solely through substitution in consumption = the lovered price of this factor makes the products in whose production it te relatively teportast cheaper relative to other products end this leads consumers to substitute then for the other products, More generelly, substitution will elso take place in production. For each product separately, producers vill have an incentive to substitute the relatively cheaper factor for others, and in general it is possible to do 20 et east to sone extent. Tee “theory of marginal productivity” is sonetines described as @ “theory of distribution". this te a misleading statement. ‘the theory of marginal prom Auctivity ab nowt analyses the factors affccting the demand for a factor of prom duction. he price of the factor depends also on conditions of ewply. The ‘tendency to speak of @ “uarginsl productivity theory of distribution” arises bex ‘emise in many problens snd contexte it ie useful to think of the supply of factors of production as given quantities —- ao perfectly inelastic. ‘This 1s particularly relevent if the problem concerns both market and nonmarket uses of factors of production, In such asses there 4s a sense in which supply condisions determine ‘only the quantity of the factors, vhlle demand conditions (sumarized in the phrase “marginal productivity") deteraine price. put note that even in this case « change tn supply — in the fixed anount of « factor == will change the grice of the factor, unless deuand is perfectly elastic. so 4t will be better in all cases to regerd 2 theory solely of the demand for factors ‘the theory of marginal proguctirsty of production and a complete theory a consisting of theory of both the demand ‘for and the supply of factors of protuction. 4 the main, the marginal productivity theory {8 a wey of organizing theake considerations that are zelevant to the denund for a factor of progustion, It has sone, but not very mich, substantive content. Tate ie reflected in the fact that: we shall be able to speak of an abstract factor of production == factor A, or By without having to speedy 4t any further. fo say that vages are equal to the value of the marginal product, for example, says relatively ttle in end of tteelf. ‘Ths function fe rather to suggest whet to look for in further analysis. Vor the value of the marginal product 1s not a single muber determined by forces outside the control of individuals or society. Tt is rather a schedule or fimetion of many variables. Tt will depend on the quality and quantity of vorkers, the quantity of capital they have to work with, the quality of the management ergant zing thet activities, the institutional structure of the mirkete in vhich they ere hired and the product gold, ote. Ta conerete applications, the basic cubstantive issue is Likely to be vit determines the marginal productivity and how the chenges ‘under consideration will afvect tt. ‘the anslysis of the demand Zor factors of protuction is closely relates ‘to the analysis of the aupply of products, and, indeed, is really only another way of locking at or organtzing the cme msterisl, In analysing the supply curve of a prouct, we are interested in tracing the effect of changes in the demand for Lt under given conditions on the factor markets. Ta consequence, we direct attention to the output of the firm or industry and take for grented the changes din the quantity of the varlous factors of production employed and in their prices fs deuant for the product and with 16 output of the product change, In @tetri bution theory, our interest centers in the factor xarkets, and so we concentrate attention on a different facet of the same adjustment by the firm. ‘To put it Aferently, the otatenent thet a fizz seeks to equate marginal factor cost to marginal value product is another wey of saying thet 1b seeks to equate marginal revenue to marginal cost rather than an a@ditionel condition on the equtlibréum of the fim. ‘fs tn the theory of supply of products, there are several different levelsokt of analysis, and the deuind curve will change as ve shift our point of view from there ‘the reactions of the firm to the reactions of en Industry. And in this ca: Se eleo @ third level thet is significant, the econouy as a yhole, since many aigterent industries may employ what in any particular problem it 1s useful to regard as a single factor of production. ‘mhe demand ourre for « factor of production by ¢ particular group of dem anders (vhich may as speciel case be a single firm) shove the maximum quantity of the factor that will be purchased by the group per unit of time st exch price of the factor, for given conditions. As in previous problens, there te cone un= certainty how it 1s best to specify the “given conditions". They clearly inelude (2) technical ImovLedge == the "state of the arte" or the production fuctions of actual and potential firms; and (2) the conditions of demand for the fins) products. The uncertainty attaches primarily to the handling of other factors of protuction, One procedure is to take as given (3) the supply curves of other Factors of production to the group of demmnlers considered. The prubles wilh (3) te thet et least for the economy as a vhole, constant supply curves for other factors may mean an inerease in the totel resources of the comunity as ve move long the denand curve for this factor in response to an inerease in ite supply. ‘The alternative is to take the "total resources" of the community, appropriately Gefined, as fixed, and thus regard changes in the supply of this factor changes in its supply relative to other factors but not in the total resources of the community. We shall for the most part beg this question since most of our Atseusse ton vould be unaffected by its resolution. Th should be noted thet the precise meaning of (2) and (3) as stated above Gepend on the partioular group of denanders considered. To a firm selling its product on a competitive market, (2) ie equivalent to holding the price of the product constant; to an industry producing a single product, it 1s equtvalent to holding the demand function for the product constant. lo a fim, (3) is equivalent ‘to holding constant the price of factors thet it buye on competitive markets, andake ‘the supply curves of other factors. In particular, tt 1s equivalent to holding constant the anount of "fixed" factors. To an industry, (3) my still be equive- lent to holding constant the price of some factors, nanely those of which the industry eo a vtole buys only s eaall pert of the total, so thet the supply curve of the fnotor to industry 48 effectively horizontal. To the economy as = whole, eopecially if thie 4e regarded as incluling the nonmarket as well. as the market sector, (3) may be equivalent to holding the quantities of other factors constant (enough tite obviously depends critically on how the meertainty sbout (3) ts resolved). Tote aiso that the aifverence between short and long- run denané curves 18 4n the prectse content of ttexe (2) and (3)- nally, the List of "other thinge” te not exhaustive for #1 problems. For muy problems, for exuuple, it will be destreble to give special consideration to closely related factors of production. Ls he indtyidual firm tn analyzing the denand for factors of production by the individual firm, ve say again start with the Améanental equetions defining ite equilibrium postition: =e 2 ye = a @ x # f(a,d,c,. (e) If there is competition on the product market, MR ¥L11 of course be equal to the price of the product or p,j 1f factor 4s purchased on competitive market, ite marginal tector cost will of course be equal to its price, For the time being we ney euppose that any factors are either purchased competitively, go that ve can replace thelr marginal factor coste by theiz prices, or are "fixed" to the firm, co that ve can regard the quantity (or maxtmm quantity) available as given. The shorter the rm, the larger the number of factors the available quantity of which are to be regarded as given, ent conver: + Tnileed, as we saw in the @iscussion of cupply, this is essentially the definition of length of run.a9 From a purely formal point of view, the denand curve for @ factor of pro- auction by an andividual firm can be derived imedietely and directly from equations (2) and (2). et the firm be selling on a competitive market, let factors a, by +++ be purchased competitively, and a, b', ... be the factors vbose quantities are ‘fixed to the firm gor the rum consSdered. ‘hen the demand curve for, say, factor Ay will be given by s a) 1 Bl, ses stand for the Fixed quantities of these factors availsble to the A 2m By my RB were firm. Now thie equation is simply a rearrangement of equstions (1) axa (2). For any given set of values of the intependent variables in (3), equations (1) ant (2) ‘can be colved to give the quantities of the various factors employed ant the quantity of product produced. This can therefore be done for every set, and the quantity of ‘A employed expressed as @ function of these variables as in (3). Xf the product market is not competitive, p, in (3) te replaced by the demand curve for x} Lf the factor mareet for 8 is uot competitive, p, 18 replaced Uy she oupply curve of D to the firm, ete. We shalt, hovever, gain insight if ve proceed more sloviy and lese formally to this final result. It is helpful to rewrite the equations (1) in the following tora: PREP, = MO» MRED, = HE, 5 @ “Ie we nave competition on both factor and product markets, these reduce to DP Pe PR Py 0) or the fantIiar equations that marginal value product of factor equal Ste marginal factor cost, in the general case, or value of the marginal product of @ factor equal-50- ‘the price of the factor, in the competitive case. Consider the first of equations (5). This shove « relation between the price of A and tte quantity: Zor each price of A it shovs the quantity of A that would have @ marginal produst whose value would be equal to thet price of A. It 4a tonpting to interpret thie as the deuand curve of the firm of A, and, indeed, the denant curve for A ts often loosely deveribed given by the value of marginal product curve for A. But this 1s strtetly correct only in one special case: that in vhich the fizm ie not free to vary the quantity of axy factor other than A, 4.e., all other factors are "fixed". For in that case, the only adjustnent the firm can ake to a change in the price of A 1s to change the quantity of A exployed: all equations other than the first in (5) become irrelevant and are replaced by equations of the form: 4 5! . the firm will move along the marginal product curve for ‘A wntil the value of the marginel product 1s equal to the ney price of A and this curve will be ite desand curve. suppose, hoverer, that not all other factors are fixed, that, for example, B cen be varied and ie purchased competitively. ypothetically, suppose the price of A to fall end the firm first to make 1ts adjustment along the marginal product curve for A, 80 St ineres 2 the exployaent of A until the marginel product falls enough to satiety the tiret equation tn (5). The rennining equations are nov n0 longer satiatied, despite the fact that they inttially were ent thet the quantity of other factors 1s, ty asmumption, the ame as inttially. ‘The reasen, of course, 4 that the marginal product of the other factors depends on the axcunt of A exployed. Sone other factors will be close substitutes for A; the marginal product of the will be reduced ty the inereased exploment of A. Other factors will tend to have thete marginal product {nerensed by increased exploynmnt of A, since ta effect there 4s loss of then per untt of A. In general, ve may expect the latter effect to dominate, as should be clear fron our earlier Aiscussion of the lav of variable proportions. he firm will therefore want to change the anount of other factors employed, reducing the employment of those vhose marginal product 1s nov less thanost tnitialiy, ant increasing the employment of the others. Bit these adjustanate will 4n tun chage the marginal productivity of A, tending to inoreage 1 for each quantity of A -- both the reduction in quantity of competitive factors and increase tn quantity of others operate tn this direction in general. The fine) position WiLL be one at which ell the equations in (5) are satisfied. At this final position ‘the price of Ais equal to the value of its marginal profuct, yet thts point 4 not on the initial value of marginal product curve. ‘he essential point 1s, of course, ‘that the marginal yroduct curve {¢ dram for fixed quantities of other factors; ‘he demand curve, in our special case, for fixed prices of variable factors. "The accompanying dsgran sumarizes the situation. The sol{é lines are Value of marginal product curve for different anounts of B (used here to stand Zor all other factors). Veaue of mar ginal product Free ofA. bab, bet, teb, bed, bed, 4 curves for A for given price oF proguct . ‘ensnd curve for A by ~ Sndividual firm gor given ~AL- ferice for ite product ena given prices of other factors. Gancity oF A per walt time Tntividuel firm: competitive product: and factor markets ‘The dashed Line 1s @ demand curve for A by the individual firs. Since competition 46 ascmed on both product and factor markets, the price of the finel product and of other variable fectors of production is the same at all points on it. But, as in such a vay as to keep equations (5) satisfied. Accordingly, che demand curve cute through the value of marginal produet52 curves, in general going through successively higher curves as the price of A falls. If demand for the product 4s not competitive, given demant conditions imply Aifeerent prices as the output varis Marginal value product diverges from value cof marginal product and is the quantity xelevant to the individual fim, With thts change in nomenclature, the preceeding @iagran can sumarize the situation, except ‘that there 1s no longer any presumption that the quantities of other factors in general vill increase as the price of A falls or that the desan! curve will pass ‘through marginal value product curves for successively higher quantities of b. The ‘reason 1s that vhtle an inerease in the quantity of A employed in response to @ decline in {ts price would in general raise the marginal physical product of given ‘quatittes of the other factors, 4 will aleo mean an increage in outyut, decline tn the price of the product, and perhaps also 0 decline in marginal revenue. This ay offeet cr nore than offset the rive in the mirginal phystenl profuct of the other fectors and so lead to a decline in the quantity of then exployed, We shall ect en analogous effect again when we conbine competitive irae and exaaine the demand curve of an industry. {Ie the market for factor A is not competitive, eo that the firm 1s a mono peontstic purchaser of A, the question how mich the firm would employ et vartous prices 1s no longer a meaningful or relevant question, since the firm affects the price by its ection ant determines the price and quintity simlteneously. The corresponting question 1s then the reaction of the firm to changes in the suEly of the factor and these changes cannot be sumarized by the ingle parancter, price of the factor, ‘hey can when the market for A 1s competitive. What vould other ‘wtee be the "demand curve" for factor A still retains significance. Tt shows the quantity that vould be purchased at various marginal factor costs. Hovever, in 0 Anterpreting {t, 1t mst be kept in mind thet @ single supply curve will tn general, have aifferent marginal factor costs for different quantities supplied, and that many @ifterent supply curves can have the sane marginal factor cost for the sane quantity supplied. (This case ts discussed nore fully belov).“5 Ia the above analysis ve have taken as cur (hypothetical) first approximation ‘the change in quantity of A with fixed quantities of other factors. This of course Implics thet even in the first reaction, the firm chmges its output. There is ‘then on additional change in output when the quantities of other factors are adjusted and of this one readjusted, Another ay of breaking doa the reaction of the firm 46 to take as the First ayproxination the change in the purchase of A that would occur Af the fiz kept ite output the sae. This 1s, ao tt vere, the pure substitution 4m production effect. If the price of A falls and output 4s kept constant, A wilL be substituted for other factors, implying in general a movenent fros the initial marginal productivity curve for Ato a lover one. At this point, all the equalities in equation (1) except the frat are eatiefied: the fim 1s producing thle output in the optimm manner, given the nev price of A. ‘he reduction in the price of A bas hovever increased the comon value of the ratios of marginal physical producte to marginal factor costes St has increased the number of units of output attainable by spending en aiditional dollar, thet ie, has reduced marginal cost. Marginal cost 1s, ‘therefore noy lover than marginal revenue, which means thet output 1s less then the optim, An expansion effect 1s therefore aided to the substitution effect. in expanding, the fra vill employ more of all factors, in general. This increase in employment of A adds to the increase due to the substitution effect. Yor other Factors, 1b offsets the initial decrease. Ae before -- eince the final position te ‘the some ~~ the final postion will tend to involve the exploysent of more of the other factors in general, but my involve the employamt of lose of close substitutes for AeSte ‘the accompanying figure shove the three curves ve have been talking about. P fe the initial point of equilibrium, and so all three ‘pass through it. The steepest (at B) stove the ancunt of A that ‘the firm vould purchase if It Xept output constant; next steapest, ‘the amount of A it would purebase at given profuet prices if 16 kept ‘the anount of other fectors em ployed constant; the flattest, ‘the snount of A it would purchase ‘at given product price and given prices for other factors. Price of A Donend cure for A Value of marginal, product curve constant output uantity of & per unit tine Individual firm: competitive product and Pector markets ‘You will find it instructive (a) to check and prove statements made about ‘the order of these curves; (b) to chow thet monopoly on the product market can change the order of these curves; (c) to translate the above in ters of pro- auction indifference curves. 2s he Competitive Industry ‘tn reacting to conditions on the product and fastor markets as they see ‘them, sndividual firme obviously change those contttions: they impose external effects on theaselvee end other firms in their own industry; and the combined ‘reactions of all firms ina single industry impose external effects on other industries. Int us first confine our attention to # single infustry. Tn response to a decline in the price of A, each initvidual fina seeks to move along its aenand curve for A, which will {nvolve expanding its output. But all individual ‘firms obviously cannot do so without changing the conditions for vhich those denand“55 curves are dram. Yor one thing, the increased output by all firms will lover ‘the price of the product and this will shift the deaand curve for A of each tne Aividual tim dovovard, since each of these 1s dram for @ fixed price of the product. This would be the only externel effect to be considered at this staze Af the industry uses no spectelized (variable) factors, i.e., if Lt employs only f suall part of the total available supply of all other (variable) factors, 50 ‘that their supply curves to the industry can be taken as essentially horizontal. ‘The final increase in the ancunt of A purchased by all firms in response to @ reduction in the price of A (to this industry slone) will be less than that shown by the sum of the denand curves for the individual firms in the industry, as shovn tn the acconpanying Plgure. The Flattest curve through Price of A Pe the eu of the denand curves Demand curve for A Sor A of the individual girms sa of industry the tndustays Uae acl steepest curve, the dommd curve for A of ‘he industry as @ vbole. Through ‘E(penand curves tor ‘A of individual each point of the denand curve of ‘fteme) (Constant output curves for A of indivieual firme) ‘the industry, there passes such a um of demand curves of individual fivms, shoving the om of the ° Quantity of A por unit time mounts the individuel firs would Competitive intustry: Supply curves ECE a) of all (variable) factors horizontal, product vere not eltered as a consequence of their increased production. The nore elastic the demand for the product of the industry, the less will tend to be ‘the Givergence between these ‘vo curves. ‘The changes in the price of the product will affect, of course, not only the anount ot A enployed, but also the amount of all other factors. As noted earlier, vith a constant price for the product, there is e presumption that the5 denand for other factors vill on the average rive with decline in the price of A. There 1s no longer any such presumption, once account is taken of the effect of the expansion of output on the ptice of the product. This can be readily sen, by taking the extreme example in which the demand for the product is perfectly inelastic. In this esse, the price of the product will fall to whatever extent 4s necessary to keep total output unchanged, snd the demand curve for A of the Aniustzy will be approxinately the enne (sn this epectal case of given other factor prices to the industry) es the sum of the constant output curves for the individual firms dram earlier. The qualification "approximately the same" is necessary because all firms in the industry need not have the sane production motions. ‘he decline in the price of A may thus affect different. tims differ ently. In consequence, the mchmged total output of the industry may conceal Aecreases in output by some firme, balanced by increases by other firme. Bat a6 ‘we sew before, these curves ixply the eubstttution of A for all other factors as ‘2 group (though not for every single one, since there may be some highly comple mentary with A), and oo reduced employment of other factors on the average. AS thie example implies, the demand curve for A of the industry will, ee shown on the figure, tend to be between the mm of the constant output curves and the ou of ‘the demand curves of the individual firms, 1ts exact position depending on the elastictty of the denand for the protuct. TF the industry uses some specialized resources, @ further external effect ‘will be on the prices of these resources. ‘The renarks in the preceding parsgraph ‘show thst ve cannot specify the direction of effect on the average. The desant for epecialized resources thet are highly competitive with A vill tend to fall with a reduction of the price of A under almost any clroutetences, and o their prices to fall. Token ty ‘tself, the reduction of the price of highly competitive factors reduces the incentive to substitute A for them, but also reduces marginal cost tnd 0 imoreases the incentive to expand output. ‘here is perhaps a presumption thet ‘the combined effect is Likely to be @ eusller increase in the employzent of A thanst Af the price of these highly competitive factors had remained unchanged. The demand for specialized resources that are highly complenentary with A will tend to rise with @ reduction in the price of A under almst any circumstances, and co their Prices to rise. ‘This tends clearly to mike for a sualler increase in the exploy- ment of A than if the price of these highly complenentary factors had remained ‘unchanged both by reducing the edvantege in substituting A for other factors and by valeting marginal cost. ‘he demand for the reaaining resources may move in either direction. The nore elastic the demud for the product, the more Likely te the demand for, and price of, these other resources to rise, in which case the aggregate effect of the changes in prices of specialized resources will be to make for @ sasller increase in the employment of A than if all resource prices other ‘than that of A had remained unchmged. On the other hand, the more inelastic the deuand for the product, the nore Likely is the demuni for, and price of, these other resources to fall, and they may fall enough to lead to a greater increase tn ‘tho employment of A than £¢ all resource prices other than that of A had remained ‘unchanged. J addition to these external pecuniary effects of the changed pattern of production stimilated by the fall in the price of A, there my, of course, also be external technical effects of the Kinds considered in the diecussion of supply curves. ‘These may operate in either direction on the exployment of A. 80 long as we restrict ourselves to the effects of the reactions within 8 single industry to the decline in the price of A, the net result vill be an in creased purchase of A snd sn incressed output of the product. ‘the effects external, to the individual fim but internal to the intustry may make these increases smiller or larger than they yould have been without the external effects but they cannot =~ aside perhaps from pathological special cases ~- convert then into decreases. For it 4s precisely the increase in output that makes the price of the protuct decline ant 6o makes expansion seen less attractive to the individual firm than At vould at the inttial prices anf the prices of other resources cannot on thee average rise except as a remult of @ generally Sncreased denant for ves viich seo means an Snereaee Sn output. Bit while Chie is true for the intustry as & hole, it need not be true for every single firm. The different firms say be using irtarent techniques of proguction and combinations of fectars. Some, for exauple, nay be using tecimiques that involve pertiowlarly hesry woo of a fector that rises sn price as a result of external effects, and for euch firms his rise tn price ray be enough to produce @ decline tn output, Some may be specially affected by external technicel effects, and 20 ons 3. Sip econcry as a whole ach of the preceding discussion applies equally in passing from each Industry considered separately to the esoneay as a whole. ach industry tn react tng to the change in the price of A Smposes external effects on ttselt and otter soaustries, Resources highly competitive with A will obviously tend to fall in price, sunt sevources Myhty coupleamatary to Ay to rise fn pricey alist no matter Row (1.0., relative to whit) their price Se nessured and what their conditions of owply are, ani there is 2ittie to wid to our previous atscuesion tout such recources. Wont, bowever, about all resources other than A in general? Obviously, ‘the fall in price of A Lg @ rise im the price of other resources relative to Ay ud hence relative to the average price of all resources, and ve are talking ‘taroughoat only about relative prices. he effect on the average price of alt resources (Sneluding A) relntive to the average price of finsl goods and services deyonds to some extent on our instal aamumptions about the source of the tncresse dn the supply of A that produrea the decline Sn {ts price (i.es, about the meaning of given conditions of supply of resources). If the increase in supply of A is ‘taken to be solely an increase in relative supply compensated ty a decrease in ‘the ewply of all other factors sufficient to keep totel resources availble umehanged in an appropriate sense, then in that same sense aggregate output will ‘be unchanged and hence the sverege price of all resources will renain unchanged“59 relative to the average price of goods and services, veh hovever meme that the leverage srice of resources other than A rises relative to the average price of Final goods and services. If the tnoreese in supply of A 1s mupposed to be a net addition to the total resources of the community, with the supply of other resources unchanged, then tt obviously peratts a greater aggregate output, Tt 1s not clear hat effect this vill have on the averege price of all resources relative to the price of final g js and services; t is clear, hovever, that the average price of ‘all resources other than A yill rise relative to the average price of final goods anf services, as in the preceding case. ‘lore as elseyhere in this section we are begging index mumber problens involved in measuring "average" price. ‘These are of the sane kind aa those considered in ‘he section on consumer demand. “he important thing throughout 16 to recognize thet we cemnot speak about changes in "prico" for the economy ap @ yhole without defining the Dae relative to witet Brice 4s neneurea. As just noted, according to at lesst one possible interpretation of "given conditions of supply of factors of production", total output mist in one sense re- nmain the sane despite the reduction in the relative price of A. Yet we say in ‘the preceding section that, 4f ve took account only of the reactions within a single industey, the decline in the price of A vould lead to an increase in output 4m ecch industry separately. Obviously there must be sone external effects that reverse this result for sone or waxy intustries. External effects via the prices of particular resources highly competitive vith or couplesentary to A may 20 60. Mare generally, however, the external effect that {8 iaportant in this conection is on the relative prices of tinal goods and services and the associated substitu ‘ton in consumption —- the effect that ve sew vorking in pure form in the cave of fixed proportions. In the preceding section, ve took account of the changes in resource prices that each industry produced ty its om reactions. But these6 ‘changes impose external effects on other industries. As we say in the previous paregraph, a decline in the price of A means thit the price of other resources in general rises relative to the price of A end also relative to the average price of ‘all resources end to the average price of final goods and services. For products produced predominantly with these other factors, this rise in their price vill mare than offset the fall in the price of A. he cost of profucing such products will ‘therefore rice and their supply curves shift to the left. This occurs for these induatetes not as a recult of thelr om reactions to the reduced price of A but because of external effects imposed on then by the reactions of other industries. ‘the output of euch industries will tend to decline, though their enployment of A ay not, for, like other industries, they have an incentive to substitute A for other factors. But the decline in output may be enough to produce also a decline in employment of A. Thus, while the demand curve for A ty every industry seperstely is negatively eloped, a curve shoving the asount of A finally employed by an industry et various prices, account being taken of all internal and external effects, need not be negatively sloped. ‘he industry may employ less A at a lover price for A, Of courve, euch cases vill for the usual reasons be exceptions. Becentially these eane coments epply 1f the increased supply of Ais taken to be a neb addition to the total resources of the comity. Th this cese, total, output can inerease so it is not impossible for every industry to increase cutput. But in general, 1f the output of those products produced with relatively little A oes not decrease, St vill tend to increase less than the output of products pro duced with relatively much A. This ts about the only change required. Te the increase in the supply of A is teken to be a net addition to the total resources of the conmmity, and 17 we suppose the supply curves of other factors to be perfectly inelastic, competition to reign throughout, and external technical effects to be absent, then the denind curve for A for the econony as & whole 16 4 value of maxginal product curve for the economy as a yhole. (You vill find 4% instructive to see why each of these qualifications {s necessary). But6 1b to not of course the sun of the value of marginal product curves for each firm separately. ‘the curves for the individual firme not only ere for given prices fof protucte, they are also for given quantities of other resources employed by ‘each firm separately. ‘he curve for the economy on the other hand takes account of shifts of resources between firms eal industry; tt Le for given quuntities of other resources to the econcuy ae s vhole. Tt shove the value of the addition to the ‘total product attainable by adding one unit of A to en unchanged quantity of all ‘other resources, ven the allocation of all resources betveen tims and industries is rearranged in the optim fashion, ‘he rearrangenente thet are possible depend of course on the ascuned conditions, in particular the afjustmt time permitted, so will be more extensive in the Jong than in the short run. For any mun, the marginal product curve for the economy will tend to be more elastic than the sum of the marginal product curves for the fires, beceuse sons renrrangenent is possible. 4nd the longer the run, the sore elastic the marginal product ewe wtll be because ‘the slder will be the range of posstble rearrangesents. Whichever jouption is made about the source of the increased swply of A, the denand curve for A for the economy as @ vhole will tend to be between the ‘oun of the demand curves for A of all individual fire, and the sum of the constant coutyut curvos for A of indtvidual fires, oo the figure on page 55/ euplies for ‘the sconomy as a vhole as well as for ex individual industry. A, Summary for competitive factor markets. ‘he denand curve Zor a factor of production for the economy as a whole re- flecks the effect of substitution tn both production and consumption. An increase in the supply of « factor and consequent decline in price would, if prices of prom ducts and other factors were unchanged, give each firm an incentive to substitute ‘mt factor for other factors in producing ite initial output and to expand ite output. The attenpt ty many firms to mike these adjustzente will, hovever, rates ‘the peices of other factors relative to product micas. his {ll raise costa of products produced with relatively Little of the noy cheaper factor relative to~62- ‘the costs of producte produced with relatively mich of the now cheaper factor, lending to cerremonding changes in the supply of these products and therehy tn thelr pric ‘als ads substitution in consmption between Sofustries to the substitution tn production within firms and industefes. ‘Thess general effects ‘will be complicated ty special effects arieing through special relations between factors in production and products in consurgtion. Pastors that are close mib= stitutes to the now chesper factor Sn profustion wili tend to fell relatively tn price, factore tnt are highly ceeplenentary, to rise én orice, with further sec ondary effects on prices of produots in the production of which these factors ere specially important. Siniierly, protects that are close eibstitutes tn consumption for the products produced with relatively mich of the noy chesper factor will tend +0 fall in price and profuete thet are close corplenents to rise in prices and Yor exch firm in the econoay separately, equiliteion requires that merging! nctor cort of the quantity of @ factor employed be equal to the marginai value product of that quantity of the factor. Yor e competitive tector market, this neans that at cach point on the econany's desand curve for a factor, the price of the tector is equal to the marginal velue product of the factor to cack fire in the cvonony separately. This te the centrel proposition tn the margins) productivity ‘theory of the demind for factors of profction. Bat as ye have seen, it is @ mach sore complex proposition that may at Sirst ampear. Different points on the demant curve Savelve not only afferent acunts of the factor in question but extensive resdjustuonts in the ongenizttion ant use of other factors, the scope of the ade Justuente dapending on the Length of run considered. he dndtvidual firm socks equality betveen marginsl velue product ané price of the factor. Tt achieves 4t by caging neviods of production and output, and so marginal velue yrodust, not by changing the prise of the factor, over which {t tas no direst control. denepsony Tt aay be vorth considering in soeyhat more dete! the case in vhich the6 Tector mrket ts not ccnpetitive. Tet us suppose that there is perfect competition onong the sellers of « particular factor service, so that @ supply curve Zor. the Zactor 4s meeningful, but that © particular fim is the sole purchaser of the factor eervice in question = the case of monopeony. As noted shove, in this case, ‘the question hoy mich the firm yould exploy at various prices 4s no longer @ mesn~ ingf question, since the firm deteraines the price and quantity simltaneously. ‘Mads case te depicted in the Price accompanying Figure, The curve WV OAT mertypothotical demand curve Isbeled the "Iypothetical denand curve" ts precisely vhat the desand curve for this actor would be 42 ‘the firm were a competitive pur= emaser of the factors and ie to be derived precisely as the demand curve ° Quantity of tector A for m factor was derived above, per unit tine ‘As vill be seen from that derivation, it shovs, for each quantity of A, the (caxtmm) anount thet the fim can afd to ite revenue per unit increase in the ‘mount of A employed. If the quantities of all other factors available to the fhm vere fixed, it yould be a marginal value product curve for A, If the quanti ‘ties of ali other factor avatlable to the firm are not fixed, the firm will vary ‘the anount of these other factors employed as it uses more or less A in such @ vay ‘86 to keep their margins} value products equell to thelr marginal, factor costs, <0 ‘the curve VW 1s no longer 2 marginal value product curve, eince the quantities of other factors are not the sae for el-points on it. Curve 8S 1s the supply curve of the factor A to the firm, It shove the naximn encunt of the fector the firm could purchase st various prices. The ordinate of any point on 83 1s therefore the average cost per unit of A to the fim Af 1 Duys the anoust given by the abscissa of thet point. qe ordinate of the curve marginal to 6S (curve Mt) gives therefore the snount that the firm would add6 to its costs per unit incres 1c in the asount of A employed ~ or the sarginal factor conte tor various anounte of A. Tt obviously pays the firm to Aire sore A so long fas the snout St thereby adds to ite receipts (the ordinate of YY) exceeds the ‘amount 4t adds to ite costs (the ordinate of Mi). he intersection of these two curves therefore gives the optimm anoint of A to employ, in this example, OB. ‘The price paid per untt 42 then the ordinate of the supply curve at T, or OP. Tote that many different prices lof the factor are consistent with the sme VW curve, snd the sane enount of Acuployed, since different supply curves can tave the sane margine2 factor costs at a particular quantity of the factor. One example is de~ pleted in the accompanying figure. ‘The factor market may fail to be competitive not because the firm is the sole purchaser of the factor but deosuse there is @ single seller. This case is essen ‘ially the seme as monopoly in the enle of @ product. ‘he seller of the factor kervices ie faced by s negatively sloped demand curve and he vill seek to equate marginal revenue with vhatever he may regard as his marginal cost. If a mmopsontstic purchaser of @ factor faces a monopolistic seller, ve have e case of bilateral monopoly. ‘he maxtmm return for the two aduopelists together 19 given by the intersection of the marginal cost curve of the monopolist ‘seller and the W curve of the preceding figures for the monopsonist buyer ~ this Se the amount of the factor that yould be used if the two monopoles combined. ‘If the bargaining between the two monopolists does not les to the use of this ‘mount of the factor, the position 1s unstable, in the sense that there is a further ‘gain thet could be gotten by merging; that is, either monopolist can afford to offer ‘the other # larger em to My his monopoly pos{tion than the value of that monopoly position to the latter - so there 4s a further desl by vhieh both can gain, Thisoe argument suggests thst there is a uniquely determined quantity under euch a bilateral ‘onogoly ~ at least if merger ie not ruled out by some non-econcate obstacles but it does not provide any means of determining how the monopoly returns vill be divided between the Gyo monopolists, and in thie respect the solution mst be regard as lergely indeterminate. One interesting special application of this nonepsony analysis has been to denonstrate the possibility that the Imposition of @ legal mintam vage higher than ‘the prevailing vege cen ratse the ancunt of lsbor employed. ‘his ie illustrated in ‘the accompanying Gagran. ‘The solid curves apply in the abst of the mintmm vege, 50 OA is the equilibris anount of labor employed et wage of Of,- Suppose a legel ininur wige of OW, 1s imposed and effectively enforced. The supply curre to the {firm 1s no longer $8, but nov becones Of, CS, since at a vage belov Gi, the fire cannot hire eny Inbor, the Marginal age rate we factor cost 48 then no longer 1 but Of © DM, whion snter sects the s omn W curve at B, so the equildbrium one explomment is 0B, larger than pre- viously, deepite rive in the wage ¥ rete fron Gi to Gig. Tn order for ths eftect to occur, i is obvious of Apter 2x ° AB plopel per ual Ess that the mintmm wage must be between ¥ and Wy, If it te above Hy Lo viit Sve the usual effect of dimintohing enployment. Tt 16 perhaps vorth noting explicitly that this case te little more than a ‘thecretical eu: sm, anf. cannot be regarded as of any grest practicel importance, partly because significant degrees of nonopsony are particularly unlikely to occur for factors of the kind affected by minima wage rates, partly because even in ouch cases there is no presumption the minimm vage rate will fall in the interval enslogeus to OF, toaa Marginal Productivity Analysis: Sone Generel Tesues 4s ve have scen, marginal productivity analysis does not provide a complete theory of the pricing of fectors of production. Tt summrizes the forces under- lying the desand for factors of production; but the price of factors depends als fon the conditions under viich they are supplied, co to complete the theory, ve shal have to analyze the forces underlying the swply curves of factors of pro- duction, Before turning to that task, hovever, it will be well to consider sous ‘general issues that have arisen in connection with the marginal productivity theory, tesues that are connected primarily with the central proposition that hhas been used in analyzing the denand for fectors of production - that factors tend to receive their marginal value product rather than with the condition of supply. 2s he Baaustion of the Product one question that arose elmost trom the outset of the marginal productivity ‘theory Le vhether there {e any assurance that payment in accordance with marginal product will exhaust the total product. May {t not be, 1t vas sata, that if each factor is paid according to its marginel product, the oun of all the payment for factors will either exceed or fall short of the total azount available to be paid? Jn this case, what happens to the difference? Wicksteed gave what vas, for a tine, the most widely accepted anever to ‘this question. We pointed out that if the production function was homogeneous of the first degree, then Buler's theorem demonstrated thet payment in accordance with marginal productivity yould exaust the totel product. let a,b, «++ de the quantities of the fectars of production, x= £ (a8) +++) be the production fune= tion. nov f(a,b,..+) 18 a bonogencous function of the t-th degree 17 e Q) FO Aa, rd, ee OF ley dy oe Zuler's theorem asserts thet for a homogeneous funetion of degree tt 2) Bp Bp edie trletd see Ie t 4s unity, this becomes G) Bad Be rds e“tm Yow the partial derivatives are precisely ‘he marginal piysical products of the various factors of production, anf they are mitiplied by the annmt of the corresponding factors. 0 each term on the left hand side de the totel payment to a tector in physical terms, if each factor gots ite pysieal mxginnl product. Jo their eum is prectaely equal to the total quantity of product available. This equality {s not altered by multiplying through by the price of the groduet, hich ie tn effect whet happens inder pertect ccapetition. ‘But this solution 1s tn a sense too good. If the production fimetion for ‘the individual frm 1s everywhere hosogensous of the firet degree, then payment tn accordance vith nargingl product will exhaust the total product regardless of ‘the proportions in vaich the factors are combined, and returns to the factors vil ‘be independent of the scale of the fia, Hirthersore, 4£ there ts any monopoly, then paynent in accordance with marginal value product vill aot exhaust the total product. Finnlly, vhtle ve have seen earlier that 1 4s reasonable to regard production fune~ homogeneous of the first degree tron @ sufficiently broad point of view, 1 does not follow from this vay of Looking at them that they are homogeneous of the tons first degree everywhere from the viewpoint of the individual firm, It te, ae st ‘were, an empirical accident if they are and tt scons rather umestiefactory to base fundamental proposition of econoaics on an empirical fact the determination of ‘Which 1s not even tn the realm of socnonios but of technology. ‘A nore satisfactory solution ts to argue thet exhaustion of the nroduct, Zar from being a necessary resultant of particular technical facts, 2 « condition of equiliteiun, Siypose a position in which 4¢ one resource omer pays ali other resources thelr marginal product he has left over more then the marginal pro- uct of the resource he ome. ‘Then all other overs of such resources have an incentive to do vhat he 1s doing and in the process eliminate the difference. Conversely if the residual is less than the marginal product of his resource, he huss an incentive to cease being a residual incoue recipient, cease his present activity, anf rent out the use of hie resource for its marginal product. the8 . romalt is that, under competstion, the individusl f{rm vill tend to operate et fx output and with @ combination of resources at vhich its production function is hosogeneous of the first degree. 2, he Role of Marginal Productivity in Positive and Normative Analysis In discussing marginal protuctivity analysis, 1¢ seems desirable to depart from our general principle of avolding normative issues. ‘The reason is thet ‘confusion between positive and normative issues 16 perhaps the baste source of nisunderstending of marginal productivity analysis and the fundaxental reason for ‘the contimal controversy about the theory. Berbaps the sisplest form this confuston takes ie in the arguzent that 1f nurginal productivity analysis ie valid, it mikes the wage rate (or rate of return to any other resource) inevitable, determined by the "lave of nature”, and not susceptible of change ty Inman action; end that this is co objectionable « result ‘that the enalyeie cannot be accepted. Of course, even if this were @ correct inference fron the analysis, {+ vould not de valié grounds for rejecting tt. We nay bemoan the fact that the world 1s round and that there is nothing ve een do sbout it; yet thet is hardly ground for rejecting it as a fact. But as is clear from the preceding analysis, it 1s not a correct inference. Indeed, the marginal productivity analysis 1s a vay of organizing the ways in which the wage rate in, say, a particular occupation can be changed ty man action: it can be raised by any action thet will reise the mergine) prouctivity of the number of employees Brea; this can be accomplished by reducing the number hired, or by raising the effictency of the workers, by increasing the efficiency of the management, 1n- creasing the onount of capital with vhich they work, ete., ete. ‘Amore sophisticated farm of the confusion ie the set of objections to mar inal. productivity analysts frequently made in elenentary textbooks - particularly in labor economics ~ namely, that the theory "assumes" perfect mobility of re- sources, perfect Inowledge of available alternatives, perfect competition, ete. Aside from the general problem raised by such statenents shout "s-o is clear that in this particular case they are entirely beside the point, 4¢ sarginal productivity 1s viewed as @ tool of positive enalyaie, a2 « means of wierstanding the forces underiying the demand for factors of production and hence of understanding vhy the prices of resources are vhat they are. Suppose labor of a particular type, for example, ts completely innobile between the South and the North, This means that these two types of labor are, in effect, Aitrerent resources, each having ts ova supply curves and the prices of these ‘to resources will be determined by the intersection of their cupply curves and their respective demnd curves, ‘The donand curve for Northern Inbor will, of course, depend on its value profuctivity, and th{s of course will depend on the price of Southern Iabor and conversely; Just as the demand for labor may depend on say, the price of land. If a mit of Northern labor is a perfect aubstitute in production for a untt of Southern lebor (1.e., many activities can be carried cut equelly vell ant at equel cost either Worth or South), demand conditions Wil1 Gictate the same price for the txo; 1f they are not perfect substitutes, ‘their prices will Giffer; and so on, Thus, marginal productivity analyais 48 Jused to deteratne the effect of imobility. Sinilarly, ignorance, tke iamcbility, will effect conditions of supply of factors of production. And we have seen how reuiily the anslysts cax cover monopolistic conditions ‘The reason theee “asoumptions" are introduced ie because they are relevant 4m Judging the normative ‘aplications of payment in accordance vith marginal product. Suppose ignorance of opportunities in, say, other furniture factories neps Laborers in one group of factories from exeking jobs in another group in which a higher yage rate ie being paid, so that essentially the suse type of dubor gets = higher wage rate in one group of factories thin In ancther, though 4m each group separately each laborer gets the value of his marginal product. Clearly, the removal of the ignorance will edd to the total product; the worker ‘who shifts from the lover paid to the higher pald ob 4e adding mare to product, in his nev employment than he subtracts fron product by giving up hie old employAs this exanple illustrates, the normative funetion of payment in eccardance vith marginal product ie to achieve effictensy in the allocation of resource: The fact that the vorker ts paid his marginal product gives him an incentive ‘t0 seek euployzent where his marginal product Ls the highest: suppose wage rates in the tvo groups of furniture factories vere equal despite the alfference in marginal product; the vorkers vould then have no incentive to ehtft vhere their marginal product 1s the highest. Similarly, it gives the buyer of zesources an incentive to use the resources best adapted to bis purpose. For exaaple, suppose there are two types of laborers, A and 3. Suppose A and B are perfect substitutes in activity Ty vhtle B has higher protuctivity in activity IE than A. Tf payment 1s in accord vith marginal productivity, B will get @ Bigher vege rate than A because its higher productivity tn activity IZ (if the amount of Bis cuffictontly Lintted 20 that all can be exployed in aquilfbriua in ectévity 11). Buployers in activity T nov have an incentive to hire A instead of B, since to get B they vould have to pay te marginal product in activity IZ. on ‘the other hand, suppose the vage of A and B ie arbitrarily made the sane, which ‘means that 1¢ camot be in accord with marginal productivity. Then employers in activity IT otill heve an incentive to get hold of B instead of A, but employ= fers in activity I have no tneontive to hire A instead of B. And, of course, in fall these cases the "incentive" iteelf depends oa the enployer's being paid sn accordance vith his margins! product, for his productivity constete preeteely in choosing the resources best adapted to his purpose, and if his reward does not depend on hoy vell he performs this task, vere ie bie incentive to do it wellt More generally, payment in accordance vith marginal product can be seen to be a means of making the rate of substitution of final products in purchase on ‘the market equal to that rate et which 1t 4s technically possible to substitute final products in production. This can be esen most readily by eliminating alioh Anternedisrjes. Suppose the sarginal product of an hour's labor, with gLren other resources 16 efter one Dushel of corn or one bushel of vheat. The ‘echnical raté of substitution 4s then one for one. Unless they sell at equal prices, the epperent rate of substitution te not one to one, ant the alternatives actually open to the consumer are falsified. So far this only requires that rates of return be proportional to marginal product, But unless the rate of return 4s equel to margins) product, the rates of substitution between market and non= narket goods vill be felettied: mypose ia the above case, the price of the Dour ‘of Isbor is one-half bushel of wheat; then to te laborer it appears that he cm get an hour of leisure ty sesrificing # balf-bushel of wheat; yet in fact ‘ie commntty sacrifices « vtole bushel of vieat. A-AAL analysis of these normative issues would involve & mush mare ex- ‘tended @iscussion, particularly on the problems raised by unagpropriable benefits ‘and unborne coste ("netghborhoed effects," divergence between private and soctel ‘profuct). But perhaps enough has been aatd to tnttcate viy pecbleae of tgnorence, mobility and degree of competition bulk large in analysing the normative tapli- cations of payment in accordance with marginal product. Ferbaps, also, enough Jus been said to indicate that the Amction of ceouring an "armropriate” alloce- ‘tion of resources is porforned ty the setting of rates per untt of resource, not of total incoues to identifiable individuals, veh depends not only on such rates but also on the anount of resources owed by individuals. 3+ Me Benes of Distrsbution ‘oe normative {sous about vhich there has been most controversy has not hoverer been the role of marginal profuctivity in achieving allocative efficiency; 4 hs rather been vhother {t also produces distoibutive justice. the marginal productivity theory hae Deon taken as a defense of the justice of the extsting Gistribution of income. Given a reasonable approximation to competitive conditions, £¢ 48 argued, marginal productivity theory shove that each man gots ‘what be produces. Clearly, a man deserves what he produces. Consequently, {%ote Jo said, the existing Gietribution of income 1s Just. one objection that can be mate to this argument 1s that private product - in eecordance with vhich en individual 1s peld ~ moy diverge from soctal product. For example, the producere of burglar's Jimys are being paid their marginal (private) product, But this objection is hardly tundenentel; 1t attacks vhat te estentisliy @ minor proulee tn the argument; for 1t merely argues that in some cases the market mecsure of product 19 not en appropriate measure of products 1% does not deny that, 1f product Le aypronriately measured, individuals should get ‘their product. ‘The baste postulate on which the argument rests 1s the ethical proposition ‘thet an individual deserves vst the resources he owe produces. Now, aside from ‘the acceptability of this proposition, it mst be recognized that tt 18 widely esd unthinkingly accepted. It 1e essential for the stability of a society thet ‘there be a set of bellefs which are unthinkingly accepted by the bulk of the society, beliers ynicn are just taxen for granted and not questioned. In my Iudguent, thie 1s or hss been one of those beliefs in our society, and the fact ‘that 1 1s co 4e part of the reason why sockety has accepted the market aysten and the associated methods of revard, he function of payment dn accordance vith marginal product may "really" be to achieve allocative efficteney. Yet it is permitted to perform this function only because 1t ie widely, {f perhaps mictakenly, Delieved that 1t profuces distributive Justice. A striking indication of how deeply this ethteal proposition ie exbedded 4m the values of our sootety is ite implicit acceptance by the most extreme op- ponents of our systen. One of Kerl Marx's chief eritictone of the capitalist aysten is hie theory of explostetion of Isbor. Labor, he says, ie exploited ‘because Isbor produces the whole product but only gets part of the product. But ‘even if there vere sone meaningful sense in vaich labor protuces the whole pro- uct, why ie the result "ba" or @ sign of “exploitation?” Tt is only 1f labor “ought” to get what 1b produces = which 1¢ the fundaxentel ethical proposition“3 stated shove, If the Fuskinien slogan "From each according to hie ability, ‘to eseh according to bis need" ts accepted (again vatving ali questions of defining the terms "ebiiity” and "need"), the entire Marician arguient leappears. To ee ablish "exploitation", one would then have to shoy not thet labor gets ‘less than it produces in some sense but rether that £t gete less than it "needs". Of course, even £f one accepts the basic ethical proposition, the Marxten ‘heory of exploitation is logtcelly fallacious. Clearly, sone part of current product is attributsble to non-human capitel. ‘he Marxien anever 1s that non ‘mman capital is the product of past labor =~ enbodied labor, as it were. But if ‘this were so (end I do not mean to tuply thet it te), the Marxien elogen would have to be rephrased: Present and past labor produce the whole product but present labor gets only pert of the produet. At most, this luplies not that present Labor is explotted but that "past labor" Sa, and a ney ethical propost= ‘ion would have to be introdueed to argue that present labor should get vhat present and pest Inbor produce. The purpose of considering this Marxian doctrine 1s not, of course, to give a complete analysis of 4% but only to show that this eriticter of the capttale tot system 49 itself based on the acceptance of the capitalist ethic. If one examines tn more detail the proposition that an individual deserves what the resources he ows = his labor pover and his nontuan cepttel <= pro duce, I think one is led to conclude that it can nelther be who2ly accepted nar wholly rejected. Yor tyo individuals in comparable circumstances and with equal opportunities, it 1s entirely reasonable because payment through the market covers only part of the use of resources and peyuent in accordance with marketable profuct {s required to achteve equality of total return. But for ‘two individuals vith mequel opportunities, the principle seems mich lese Foasonable. One man is born blind, another with Mis eight; is dt “Just” that ‘the former receive less than the letter because his productivity 18 anallert The ditficulty is that 4t is hard to sce any other principle to exply. Theothe fundamental "injustice" ts the original distribution of resources - the fact ‘that one vas born blind and the other not. Yet it de clear thet in such eases ‘we do not in fact apply the principle of payment in accordance with product. ‘Tt is sonetines thought that the principle applies more fully to pay- ment to Iumen resources than to payment for the use on non-human resources; that on indtvidual "deserves" what he produces by MMs labor or the capital he hinself produces but not whet the capital he bes inherited produces, As the preceding example suggests, this distinction is largely false and irrelevant. If ¢ man deserves vhat he profuces ty Ms lebor, that means he is entitled to consume his product as he wishes to provided he does not interfere with others. If he may use It in riotous living but may not give 4t to his son te he getting what Ms Isbor produces? Mr. Jones inher‘ts wealth from Ms father; Mr. Gaith ine herite rere and highly paié physical or mental ability. Altematively, Mrs Jones uses the proceeds of his labor to give hie eon a technical education, whieh will enhance hie son's earning povers Me. Saith uses the procesda of his lsbor to by ‘2 business for his son, vhich vill enhance his son's earning pover by the sane amount. Wharein 1s the difference? This te © superficial and incomplete Atscussion of complicated and eubtle Asmues. Its purpose 1s not to present any complete or satisfactory analysis but rather to shoy that the marginal productivity enalyeie of the determination of rates of retum to resources does not have any unique ethical implications. Ae- ceptance of this analysis in no vay camits one to an acceptance of the extsting @isteibution of tnccee ‘the right or the appropriate Qictribution == or, for ‘that matter, to ite rejection.ce The Factors of Production Our discussion of denand for factore of production vas in highly abstract ‘terns; ve aid not consider the specific character of the factors of production or give then nance. The reason 1s that on the demand side, there seens no empirical clatsification of factare that has euch special importance a2 to ae serve being singled out; the classification that ie usefd will vary from problen to problem. On the demand side, the chief consideration in class! fying factors is, of course, oubstttution in production. A single factor consists of unite ‘that are regarded ac perfect substitutes in production; different factors of unite that ere aot perfect substitutes. For gone problems, it will be desirable to separate out: muy different factors of production; for others, only @ fev. Tt has traditionally been supposed that conditions of supply give @ more substantial and empirically significant basis for distinguishing snong factors of production in specific terms. The classical economists aletingulshed three main factors of production: land, capital, and Isbor. Land they regarded aa a permanent non-reproductble resource fixed in amount, whose supply vas therefore perfectly inelastic to the econcay as a vhole. Capital, they regarded as @ reproductble resource, the aucunt of which could be altered through delsberate productive action, so ite eupply was not perfectly inelastic, Infeed, in the ain, they tended to regard it as highly elastic. Tabor, Like capital, they re garded as reprofuctble and expansible, and, indeed, as supplied to the econosy in the long-run st constant cost, yet to be distingutehed from capstal. because of Ste dual atatue 4 productive resource and an ultimate consumer. This particular tripartite Aivision vas doubtless a consequence of the perticuler social problens that vere inportant at the time the claseteal theory yas developed and the social structure in which the industrial revolution occurred in England. There may still be sone problent for which it is important ‘to @ictinguish land trom other resources; but for most problems 1t hardly sesns important to do co. In most contexts nov important, lend, in any economically16 relevant sense, is indistinguichable fron other forms of capital. Productive power of the soil can be produced at a cost by drainage, fertilization, and the ke, and ts clearly not permanent. land rent, even in the custommy meaning of the tera, hae becone © mich amiller fraction of total income in advanced countries in the course of time. From a broad vlovpoint, there te mich to be sald for regarding all sowees of productive pover as capital. Much of the productive pover of what ve call Isbor 46 clearly the product of deliberate investment, and is produced in the same sense as machinery or buildings. Human productive power Se mubstituteble for non-tumen productive powers and can be produced in place of the latter at a cost. Indeed, one of the striking fontures of capitalist development te the ‘tendency for a larger end larger fraction of total investment to take the form of man capital. What 16 deotenated "property" incone is in general « sasller fraction of total incone - despite the mch greater absolute anount of piysical capttal ~ the more advanced the scctety. Tt 4¢ a snaller fraction in the United Btstes, for example, than in Dura or Infia, probably also than in Prance or Great Bettain, and probably also tn the United States today than « tnndred youre seo. ven though we recognize that all sources of productive services oan be regarded ac capital, our social and polisical institutions make 4t decirahle to recognize that there is ax isportant atetinetion for many probleas between tvo ‘road cotegories of capétel ~ Inman capital acd nonchen capital. Ke can ear plore the significance of this distinction ty exantning Marshall's @lecussion of the qpectel "peculiarities" of labor which in his view justify aistingulehing tt from other factors. He lists five peculiarities 1, "the vorker sells hie york, but retaine capital in himself." 2. "he seller of lebor mst deliver 1 hiawelf." "Labor is perishable." “the sellere of tt are often at a Gisadventage in bergaining.”oT 5+ A "great length of tine /Ts 7 required for providing additional supplios of specialized ability.” ‘As Marsball recognizes, the first two of these peculierities stend on a rather different footing then the others. Labor ie perishable in the sense ‘that the depreciation of the source of labor services (the Inan being) depends primarily on tine, rather than rate of use, so 4¢ today's labor services are not ‘used they cannot very readily be stored, and there 42 mt a correspondingly larger snout available toucrrov. But thts is equally true of mch non-human ‘capital - the services of @ Uridge or a road or & machine that deteriorates Primarily vith tine, or, economically speaking, of an automsbtile, vhose physical characteristics can be preserved but whose economic value cannot beemte of cbealescence. Again, the bargaining disaivantege 12 by no means alvaye on thé eide of Anbor, ae Marshell points out, end as experience bas amply denonstrated since. Insofar ac there Ls any wystemitic difference on this score, 1t vould seem to bbe en indirect effect of iten (1). Since uon-mmen capital can be bought end sold, it 1s easter to torroy on such cepital than 1t 1s to borrow on prospective earning pover, and it is possible to get funds by selling come of it, whereas ‘this 18 not possible with human capital. More generally, a "bargaining" problen of any kind arises only when the market ie not competitive; and indeed, strictly ‘speaking, only when it 1s competitive on neither the selling nor the buying side. But then the bargaining advantage depends on which party is the monopoliet, oF Af both are, on their relative monopoly poer, and it is hard to see that this, Gepends intimately on whether the resource in question is or te not labor. Again, item (5) 1s et most © question of degree. A great length of time is required for other kinds of capitel: the Suez and Panama Cansis; the invest ments involved in the early stages of the radio, aviation, and television in dustries come readily to mind. ‘Teens (2) and (2) are on a different footing, since they derive from theaT bbastc institutional cheracter ct our society. ‘these pecullertties vould ate- sppear only in a slave soctety and there only for the slaves. ‘The fact thet ‘muna capital sources camot in cur soctety be tought or sold means, se vas noted shove, that tnan capital does not provide as good & reserve against energencies es nonlnsan eapttals in consequence, the larger the fraction of axy given total ncoue that cones fron mman capital, the greater we should expect to be the deeize to save. Tn the eecond place, this fect reduces the scope of market forces in investment in Inman capital, ‘The individual vho invests in a machine can om the machine and so be sure that he gets the return from his investaent ‘The individual who invests in another intridual cannot get this kind of aseur~ ance. fd individuals have incentives to invest sn theaselves or thes pro gory thet they do not have to dnvast in machines. Go there may readily be either underinvestaent or overinvestuent in lnman relative to nonehusan cepital. Finally, the fact that humin capital sources cannot be bought and cold te ‘the basic reson for Marshall's second peculiarity: it is only for this reason that the seller of Inbor mst deliver it himself, Bit this mess that non Pecuniary considerations become relevant to the use of Iusan capital in @ ‘way that they do not for non-tuan capital. the omer of land, for example, has no reason to be concenned vhether the land is used in a way that de “pleasant” cor “umplessent"; or the oer of a horse, vhether the horse 1s used in work that At "enjoys" or does not enjoy", provided both types of vork involve the same effect on the horse's subsequant productivity. he omer of labar-pover, on ‘the other hand, does have reason to be concerned. He i required, as 1t were, ‘to make a tevin contract: his sale of labor-pover 1s tiedein with the "purchase" of the condttione of work, the pleasantness of the task, ete., ete. ‘These special considerations applying to human capital affect its supply in ways that deserve further consideration, co ve shall tura to a consideration of the cupply of Inbar in general in the short and long run, and then of the swyply of labor in different cecupations. Sim{lar consideration is not required for the other factors.“1% She Supply of Labor as a Whale ator 1s, of course, not homogeneous; an hour of labor of a attch-dteger is not equal to an hour of Iabor of an aixplane pilot. Yet, as always, we can ‘think of constructing a supply curve for Isbor in general by taking for granted sone structure of wage rates, and adopting some convention for adding together Gigferent Kinds of labor. For example, ve may define our assumed structure of wage rates in terms of fixed ratios of wages, and then convert actual houre of Inbor énto “equivalent” hours by using these ratios. For example, if ve suppose ‘the wage rate of the pilot fixed at 10 times the vage rate of the Aiteh-digeer, ‘ye can rogard one hour of the pilot's labor ae equivalent to 10 ditendigser hows. In this vay, we can conseive of the total mmber of equivalent houre of ubor supplied as 2 function of sone index mumber of the structure of wage rates, say the rate for the ditchedigger, recognizing thet at each such rate, the total owply consists in fact of so many houra of Aiteh-digger's labor, 60 mich of pilot's labor, etc. And, as always, in folloving this procedure, we are not supposing that the structure of relative vage rates 1e in fact determined outside ‘the economic system or independent of the level of wage rates; ve are simply Gividing up our problens and considering then one by one. Tt ceens desirable to distinguish between two Kinds of supply curves of labor in general: the supply of Iabor for a given gepulation of given capacities - ‘the short run supply of labor; and the supply of Labor without such restrictions = ‘the long-run supply of Isbor. The second clearly involves # "theory" of popula ‘sons 1. Mie Short-Run Supply of Labor oar given conditions obviously mean that the short-run supply of Labor for all purposes 1s perfectly inelastic: 2: hours times the mmber of people is the available datiy supply of labor 1f ve neglect the corrections for diz- ferent qualities of labor. But clearly, the problem that we are interested in Le the cupply of Isbor, not for ell purposes, but for use through the market.fo. So the probles ve are concerned with 1s essentially the factore that deternine ‘the fraction of the totel Isbor power that i offered for sale on the market. ‘a our modem society, this fraction 4s relatively mall, 0 there 46 cone siderable room for variation in 1t- Something less than bal? the total population te el fied as “in the Iahor force", and these infividunis devote only # minor part of their total tine to market activities perhaps one-quarter. Moreover ‘the fraction has undoubtedly varied considerably over tine and from country to county. Perhaps the ost widely accepted hypothesis sbout the short-run supply curve of Isbor te thet 4t is backyard bending above some vage rate, a5 in the ecccapanying figure. Bach point on this conve Ss to be tuterpeted ax storing toe gs macimm quintity offered at the given ewe price, which 16 why the negatively sloped seguent 1s said to be "backward bending” rather than “forvard-failing". A variety of empirical evidence points to this con Flow of labor services per unit clusion. Th the first place, as the real of tine. Stort-run supply curve of labor. wage rate hae increased eecularly over loag pertotis of tine in advanced comtrios, the average muber of hours a week bts tended to decline, and the fraction of children in the lebor must to decrease. ‘he fraction of women has not behaved co eystenatically but hes probably tncressed. Yet all sn all, 4£ much observations over a long period of tite were regarded as eing on the swply-curve, they vould produce backward bending seguent. Addi ‘ional evidence is furntehed by experience in underdeveloped coustries, where it seems to be common experience that beyond @ fairly lov level, an increase in wage rete per hour will reduce the mumber of hours worked; that the natives act as if they vonted a certain sum of money almost regardlest bow long they have to vork for ity ££ they can get that sum in fever houre they will werk fever ours.a The theoretical explanation offered for the backward bending soguent of ‘the supply curve is that @ rise in the real vege rete arising fron an increased Aesand for leder has tuo effects tt makes Leisure nore expensive, eiuce the cost of an hour of Leisure 1s the vege that could be ented in that hour = thie ie the substitution effect and by itself vould tend to raise the musber of hours worked; Sf the individual were to work the gene number of houre, it increases his eal income viich vould lead htm to want to purehase nore of various kinds of goods, including leisure - thie is the inooae effect and ty itself would tend to yeduce the number of hours vorked unless leisure is an inferior good. The argue ment then is that beyond sone point the incae effect dominates the subetitutton effect. It shove up in people working fever hours, in the withdrawal of supple- mentary vorkers (children, wives, etc.) fron the labor force, ete. This vay of putting 1 also makes it clear that mich depends on the relative value attached to gools purchased with money through the marltet, and to goods that can be acquired ‘hrough nonmarket activity. In the primitive society, the inttiel lov wege rate ft whlch the Income effect Dotones dostnant reflects the lack of familiarity vith sarket goods, the Linited range of tastes. As tastes develope, and imovletge spreads, the potnt at which the income efvect dominates tende to ri Te 4s sometimes objected to an analysis Like the sbove that individuals cannot determine for theuseives the muber of houre they work; that this Ls an ‘Snetitutional datm Which the individual mst take or leave. hie objection te imost entirely specious, ih the first place, ve have seen that mich of the ade Justnent may take the form of the fraction of the people in the labor farce. In ‘the second place, even at any given tine, @ particular individusl hes sone lee ‘way. He ean work overtine or not, take off more or less tine during the year, choose the Kind of occupation or employer that offers the mumber of hours of Work he wants, ete. But neither of these 4s the basic fallacy in the objection. Mee (eportant point te that the individual 4a Like the perfect competitor: to eon individual separately, the nunber of hours of work per week my be fixed;80 yet the level at wh{ch {t Sa fixed ie the result of the choices of the individuals 2s a group. If, at any manent, thie level of hours is, say, larger than on the average people prefer st the given vage rate, this means thet any exployer who ‘makes then shorter, who adjusts them to the workers! preferences, will malte ex ployment with him more ettractive than employment with others, so can attract ‘the better people, of attract people at a lover wage rate, Beployers thus have an incentive to adjust working conditions and hours to the preferences of the workers, (In our earlier terminology, bectuse of the tie-in character of the ‘twaneactiton, exployers are sellers of conditions of vork es vell ss buyers of iabor.) Competition in this vay does permit, in effect, tniividuals to determine for thonselves the mumber of hours they york. Authough the supply curve under @iscussion 12 @ short-run curve in the sense that £t holde population constant ve have been talking in teras of the effect of alternative levels of real wage-rates each of viich Le regarded as perssnent, Lies, ae expected to contimie. Clearly, the reaction to a higher vage rate ex: pected to be temporary snd then to revert to a lover level will tend to be very Astferent than to a higher vege rete expected to be permanent. ‘The texporarily higher wage rate would seem nore likely to bring forth an increased quantity of isbor fron a fixed population than a permanently higher one, since thare would bbe strong tenptation to take advantage of the opportunity while 1t lasts and to ‘buy the leseure later. ‘in dnteresting case in point te the experience in the United States during World War IT, when both the fraction of the population im the labor force ant the average mmber of hours worked por veek were substantially higher then during the presver period. At first glance, it seens that this Increase cannot reflect response to a higher real vage rate expected to be temporary: money vages rose sharply but so a4€ prices, both openly end indirectly through deterioration in ‘the quiltty of products, co that average money wages per unit of time atvided ty fan intex of prices of consumer goods corrected for quality deterioration may not8 have risen at all and my even have fallen. Sone econonists have rationalized Unis apparent conflict between a constant real vage and an increased quantity of labor supplied by introducing the notion of a "money {iluston", namely thet sup Pliers of labor react to nominal money wage rates, not to real vage rates, that ‘hey would behave differently if, say, ali nominal prices and vages were doubled. Tt te not, hovever, necessary to introduce this deus ex machine of @ money {21u sion to explain thts phenomenon. Te can readily be rationalized on the grounds ‘that the apparent fatiure of real wages to rise is Steel? an {1lusion for two reasons. First, many additional persons vbo entered the labor market yould not have been hired previously at the prevailing real vage rate; the real wage rate they could get increased even though average wage rates Gia noty indeed it Ss possible for the real vage rate to have increased in the relevant sense for every Andividual separately yet for the average to have remained unchanged.* To illustrate this possibility, suppose there 1s no variation possible in the munber of hours vorkea by a laburer if te works, the wage rate st wnich 1avor of type A can initially get eaployment is $1 an hour, Isbor of type B, $50 an hours there are 50 laborers of type A snd 50 of type Bj the laborers of type A are villing to vork at $i an hour; lebcrera of type B are Unwilling to work at $.50 an hour. Initially then only labor of type A vill be working and the aver~ fage wage rate will be #1 an hour. Let the (real) wage rate offered for labor of type A go up to $1.25 an hour, and for isbor of type B, eo $.[5 an hour, suppose ‘Ghat at these wage retes, laborers of doth types are villing to york, asd that, Doth vork the same number of hours. The average wage rete vill still be $1 an hour, yet the vage rate thet 1s relevant to the supply of lnbor has risen for every vorker seperately. Second, people may very vell have thought that the rise in prises of consuer sods during the var ves tenporary and that after the var prices would return to ‘thets pre-war level. Any part of their vages saved should be deflated ty the ompected postevar, not var-tine, price level; but if this vere done it vould be seen that reat vages, as evaluated by their recipients, vere higher than vould be indicated by deflating by current prices alone. ‘This second force 1s especially important, if, as has been argued, part of the increase in labor supplied 1s to ‘take advantage of a tenporary opportunity. For this vould mean that 4t vould be lanned to cave an stnormally large part of any increase in income, viich wouldBle make the expected future price level pertioulerly important. ‘This interpretation ie indirectly supported by e muber of facts, in particular, the sbnormelly high fraction of income enved during the war perio’, and the extent to which such savings vere accumlated in the form of assets fixed in noainal value (government bonds, cach, ete), rather than equity securities cr real goods. Of course, th= expectations about the future price level were, in the event, Geappointed, but ‘a mistaken prediction of the future ie dn a very different class than an illusion Shout the prese 2. The Long-ftun Supply of Labor Ie we turn to the problen of the long-run supply of labor, ve mst analyze ‘the effect of the real wage rate on the size of the population anf the quality anf kills 1 possesses. We need, that te, @ theory of population and a theory of investment in the lumen agent. It ie clear thet these tvo are not unrelated: adastionel shor power cnn be produced either ty increasing the mumber of Iaborers or ty investing more capital in exch Isborer. For stxplictty, ve shell phrase the following dlscussion in terme of the elze of the population, though mich of 1 also applies to investwent in the humin agent. "To begin with, the theory of population vas regarded as an eaential ele- ment of econonie theory, and the Melthusien theory of population wes a cornerstone of classical econcate theory. In its crufest form, the Malthusion doctrine was ‘mt Leder was a form of capital that, Like other capital, could be produced at a cont; thet 1t vas produced under conditions of constant cost, the level of this constant cost being the mint etandard of Living consistent with preservation: It the wage rate provided standard of Living above this level, marriages vould ‘end to occur earlier, the birth vate to rise, the death rate to fall, end the ‘population vould tend to increase, and conversely. In this form, the theory Leads to a perfectly elastic long run supply curve of labor as in the accompany85. ing Gingram, where OW ie the vage rate real wage that provides the minima etendara of rate aaving. ‘Even in thie crude form, the ‘thecny 1 consistent with mich observed W 5 evidence, both that avaleble to Malthus a Wow of 1abor and experience eince his time, Same ex services per unit of tine treme examples are rurnished by the Phillipines and Puerto Rico The large suourt of oxpital invested in the Phillspinee by the United States over the period of a half-century has been accompanied by an epproxinate tripling of the population with ttle or no change in the average stand- ard of Living. Simtlerly, @ major effect of increased U.S. aseistance to Puerto Rico, especially since 1933, has bean every repid rise in population. And nunerous other exmples could be etted. ‘At the sane tine, 4? OW 15 interpreted as essentially a technologically Aotermined datum, the experience of ost countries in the Western vorld contradicts ‘the crude Malthusian theory. Tn such countries, the real yage has ricen drastically in the last century and a half. ‘rue, population roae also but ty nothing 1ike the extent that vould have been required to wipe out the gain in average real income. This apparent contradiction of the Malthustan theory led to Sts rejection ‘y economists and, indeed, essentially to the exclusion of population theory 2ron econenics. Population, it was sald, depends primarily on host of non-economte considerations which are not within our competence or field of interest. For our purposes, ye shall take population for granted, and leave the explanation of population change to dencgrephars, scctologiste, and the like. More recently, econonists have renewed their interest in population theory fant have become again concerned with reintegrating the theory of population vith economic theory - @ development that 1a to be encouraged.86 one vay of working tovard a theory of population that 1s consistent with ewerience in the Western world and et the sme time is coherent with eeononic theory as e vhole i¢ to re-examine the Malthuslen theory and interpret 1¢ in a more sophisticated fashion. Instead of taking the essence of the Melthustsn ‘theory to be the existence of a technologically determined cost of production of ‘unan beinge, we oan regard ite essence as being the notion that the production of Inman beings is to be regarded as if 1t vere 2 deliberative economic choice Gcternined by the balancing of returns anf costs. From this point of view, the children are to be regarded ine dual role: they are a consumption good, a way of spending one's incone to acquire satisfaction, an alternative to purchasing automobiles or donestic service or cther goods; they are a capitel goot produced ‘ty economic activity, en alteretive to producing machines, or houses, or the tke. Viewed ae a consumption good, the ancunt produced will be detenained by ‘the relative cost of children versus other consumer goods, the incom: avaliable for all uses, and the tastes ant preferences of the individuals in question. on economic forces enter the picture primarily in determining these tastes and pref- erences. Viewed as a capital good, the sncunt produced will be detersined by the retums that this capital good i expected to earn relative to other capital goods, ‘ant the relative costs of producing this ant alternative capital, goods. A major Aifterence between this and other capital goois, of course, 16 the possibility of appropriating the returns ty the individual who makes the initial capitel investe ment. Of course the fact thut children are, in this sense, a Joint product means ‘that the tyo sets of consideration need to be coubined: the returns from the children ae capital goode may be taken ac reducing their costs as consumer goods. Were 1% not for this factor, tt 1s pretty clear that gross unterinvestaent ta Jaman capital vould be almost inevitable in a free society. From thie broader point of Toy, OW a cur earlier diagram ts not to be re garded as @ technologically determined datum but as a rather complex resultant of81 ‘the factors Just discussed - a phenomenon that was alreedy emphasized in Malthus's ‘Uime in the deseription of OW ee a “conventional” minimum and the exphasie on the possibility of raising tt by altering people's tastes and values. ‘Rong these Lines, vo vould argue thet the failure of population to in crease in the Western World as fast as the crufe Naltiusien theory vould suggest nay have reflected eimply a rise in the costs of children relative to their ree ‘cur, end need not even have involved a change of tastes. A number of factors ‘hat presummbly operated in this Atrection come to mind: (1) The cost of raising children is clearly greater in the city than in rurel areas and economic develop ‘ent in the Western World involved extensive shifte to cities. (2) Returns fron children as capital goods are also lover in the city than in the comtry because ‘they are in general less valuable at early ages and, moreover, the mores in the city are such that they are Likely to cease contributing the returns from thelr productive use to the fantly at an earlier age. (3) The loonening tantly tes ‘thst case a5 a concomitant of infustrialization made the children lose valueble a5 8 means of providing memploynent end old-age socurity. (K) With growing vel Income, the spect of children as consumer's goods becaue nore important ‘than as factors of production = that is, the services yielded by children as a consuser's goot are a swerfor good. Dut thie meant sending children to school, longer and keeping then out of the labor mrket longer, so reduced the positive return to parents fron children and increased the cost involved, so made chiléren more expensive relative to other consumer goods. ‘This Liet is not intended to be exhaustive, but rather to be suggestive. Clearly cone comterbelancing {tens need to be included as vel. The modified Maltiustan doctrine may be consistent not only with the hie~ ‘Vorical developments in the Westera World but also with many currently observed phenomena. Yor example, the higher birtherete in the country then in the eity is clearly consistent with the considerations cited abore, Indeed, fran thie point of ‘View, the tendency for there to be a net migration from the country to the eity sn88. ‘the United States can be interpreted very differently than it generally is. Tt is generally interpreted as reflecting a divequlltbriue position in the process of being corrected, but with 90 such friction that the corrective process proceeds slowly or "too slowly", so the return to the farmer te on the average belo ite long run equilibrium value relative to the return to the efty dveller. he al- ternative interpretation suggested by the above 1s that rural arene have a cone parative advantage in the production of human capital ae vell as of food; that people in rural areas are involved, as {t vere, in two industries thet are pur sued Jointly ~ the production of food and of munan capital, and thet they engage 4m net exports of both to the city. On this interpretation, the net lov of pope lation from country to ety 19 no evidence of Aisequilibrium but of equtlitriun and part of the returns to rural faniites are the returns they get either in pecuntary or nonpecuntary form from their children. Another observed phenosenon that zay fit in ith this analysis 1e the strong tendency for the muber of ch{ldven produced per fantly to be eualler in ‘nigher" socio-economic classes than in “loyer" socio-economic classes (anong prom fessional and business people, for example, than anong unckilled workers). Yet ses for the not clear viether there 16 a tenfency within socio-economic cls uber of children to be lover the Righer the income. Items (3) and (b) above tn Atcate one way in watch these phenomena can de explained. Because of aifferent ‘taste and opportunities, the relative costs of children are diferent in aitver~ ext socio-economic classes. Perhaps the major factor ie that in the higher classes, ‘the child ts Likely to stay in school longer and, of great importance, to get a Kind of education which must be privately peld for, whereas in the lover classes, it is more Likely to be publicly paid gor or earned ty the child himself. thus children are more expensive relative to other consumer goods the higher the soclo- ‘economic classes. But these factors may not operate within socio-economic classes, 150 tt yould not be surprising to find that the higher the income within such « class, the larger the musber of children.85 Again, indirect evidence for euch an interpretation 4s provided by the relation betyeen the birth rate and general economic conditions, and by the tects of special subsidies provided by state action for children. Both litler nl Massolini introduced euch subsidies, and various family allowance schemes, example, the current French sche, involve such a subsidy. There seems sone evidence that euch echeues have in fact had'a eignificant effect on the rate of population grovth. This enalysie can by no means be regarded a6 vell established - or even well defined. But i$ does seem one of the more promising directions in which fan econonie theory of population is capable of being developed.‘The Supply of Labor in Different Oso. In Uscusaing the eupply curve of Labor in general, ve have taken for granted the structure of wage rates for labor of different Kinds -- relative wages in Gifterent occupations. his structure of wages is itself determined by the relative Geman for and supply of labor of different kinds. The reason ve have been putting it aside and are sble to analyze it separately is because the major forces detenaining the supply curvee of Labor in particular cecupations can be regarded as largely, though of couree not entirely, independent of those deter= mining the totel oupply of labor. At may given tine, there vill exist sone structure of relative wage rates (or average earnings) in Aifferent occupations. It 1s useful to regard this structure a2 the result of three kinds of forces or phenomena producing Gifter= entials between vege rates in different cocupations: Ls Factors gther than wage rates thet affect the attractiveness of ais fereul qvowalious Vo iallviduade 1n s position to choose among tem. ven if ‘there vere perfect competition, perfect and costlese mobility, and all members of ‘the population hnd identical abilities, money vage rates in different occupations would by no means be cqual. Sone occupations will be less attractive then others and vill therefore have to offer a bigher vage than others if they are to attract people to then. Given differences in tastes, the precios set of differentials ‘tut will arise In this vay depends not only on the characteristics of the occu ations, but ales on the conditions of demand. For if the demnd for an occu petion is relatively email, it may be possible to staff it entirely with people ‘who regard Lt ae nore attractive than other occupations in vhich case the wage rate would, on this score alone, be relatively lov; 1f, on the other hand, the Gonand is relatively large, it can be met only by attracting people into the cecupation who regard other occupations as more attractive, in which case the wage rate vould have to be relatively igh. Differentials in vage rates which arise from this set of forces may be termed equalizing differences.1 2. Factors thet produce nomeompeting grouge. Por a variety of reasons, not @11 people are in @ position to choose freely -~ even at sone one time during ‘their Lifetime -- mong occupations. he existence of such bamciers to the stat fing of particular occupations produces a series of ertly sheltered, though not entirely ureleted, markete, and inklbite the operation of the farces aiscusced 4n the preceding item, Difverences in naturel ab{lity con be classified under ‘this heading, although they could perhaps also be classified under the preceding fone. Differentials in wage rates arising fron this set of forces may be tered differences arising from non-competing groups. 3+ Tacomplete sdjusteent to changes in demand or supply, ‘the imediste effect on vage rates of any change in the denand for or supply of Inbar of various kinds may be very different from ite ultinate effect. And this ts a market in which it may take @ long tine for the ultinate effect to be felt - ‘for the immediate effect to produce reactions that will lead to a ney equile Aurium. At any tine, therefore, some part of the differences in vege rates may be regarded as attributable to ineompletent of adjusteent. Of course, whet cones under this heading depends on one's viewpoint, on the conditions ‘that are being held constant for the purpose in hand, eines hy "adjustment", We mean adjustuent to sone given eet of condition Ht these elven set of conditions define market denant ant supply curves, the existing position involves AAAI adjusteent to then, and nothing cones under this heeding. The longer the run, which neans the nerrover snd more ultinate the set of conditions taken es given, the more conse wider this heading. Difterentiaie in vege rates erietng ‘Som incompleteness of adjustuent may be termed traneitional aiszerences. 1. Equalizing differences in wage rates {To simplify the discussion of the supply of Isbor tn different occu ations, let us concentrate on two particular occupations, say A end B. We cen, ‘then sumarize the conditions of supply for thece occupations aa in the ac-92" companying figure. ‘The vertical axis shows ‘the vege rate in A relative to the vage rate in B., both being expressed tn sone comon ead conventent fora, say per hour. ‘The hort zontel ate shov the mumher of man-hours supplied in A relative to the musber tn Be ‘Tae curve then shove the maxim relative munbor of man-houre that would be supplied at various relative vage rates. ‘This wethod of cumarizing supply conditions is not, of course, perfectly general, and implies soasthing about the conditions of supply. for 1t might be ‘that the relative muaber of man-hours supplied depends not only on the relative wage aye but ales on the absolute wage rates ~ for example, thet the relative supply ‘would de diferent at wage rates of $3 in A and $1.50 in B than at wage reges of $6 in A and $3 in B, However, this kind of effect ie not something ve are going to be ole to cay much sbout, snd ite neglect is more than compensated tor by the convent~ fence of the above mode of sumerizing supply conditions, Of course, the swply curve is only valid for given "other" conditions; in perticular, for given alternee ‘ive eeploynent opportunttie ‘If all individuals hed identical tastes and abilities, they would, given the sane information, evaluete identically the relative uerite of &ifterent occupations. The reoult vould be tht a supply curve Like that drawn above youla be horizontal: ‘there would be sone relative wage rate that would be regarded by all as making the ‘wo occupations equally attractives at any Mgher relative wage rate ell vould go into As at any lover relative vage rate, all yould go into 3. Differences in tastes, sbilities, or information about the two occupations will lead to differences axon individuals in the relative vage rates regarded as making the tvo occupetions ‘equally attractive and will introduce @ elope into the curve as in the sbove Plgure. We can organize our discussion most conveniently by classifying the factors<9 affecting the supply curve into three categories: (a) thoee that determine the relative pecuntery attractiveness of the tyo occupations; (b) the vardebility of income in the tyo occupations; (c) non-pecuniary differences auong the occupations. ‘A aajor reason for this particular breakdown is thet the factors in the firet cate: gory affect all infividunis (at least of equal ability) alike and eo stould affect, mainly the height of the supply curve - they are almost the only factors that vould have to be talven into account in a “slave” society and theis counterparts are zele- vant in raving the supply curve of the services of non-human capital for one use or enother. ‘The second snd third categories introduce the factors that becone Smport= lant because of the peculiarities attached to hunan eapttel. (a) Factors capable of actuarial evaluation consider a slave omer deciding yhether to spectalize and train his slaves for pursuing occupation A or cecupation B, his decision may not, of course, be Ssrevocable - an infividual trained for A my be sble at a later date to enift to 5, ‘but generally only at a considersble cost. In making Mis decision, the slave omer would vant to imov mich nore than the wage rate per Bour in the tvo cecupations: ‘A for example might be seasonal, B not seasonal, vhich might make the expected suber of hours of york por year lover in A than in Bj A might te sore affected by eyelica movements tha B, 20 the expected umber of years of vork lover for this reason in A then in By A might be an occupation requiring grest physical strength 60 that the expected muber of years during which an individual could be employed in A might be lover than in B, which might be a sedentary occupation; A may require fa longer period of training; ant so on and on. ‘Tho effect of ali euch factors cen be summrized in the expected net returns ‘from each oceupation for any given wage rate and for each age of the vorker, as in ‘he accompanying Alagran.Beected annual net returns Af occupation A ie chosen Age ‘The net return for any occupation and year depends, of course, on precisely what are regarded as occupational expenses and so deducted fron gross returns. A Literal slave omer would regard the cost of feeding, housing, and clothing the slaves as en occupational expense; he would be interested only in the excess of earnings over ‘hie oun Thanks to the dual nature of human beings in ou society ~ 8 factor of production and the ultinate consuser to satisty whose vants prouctiion is carried on ~ 4% 4s tepossible or nearly so to distinguish the part of a man's consumption ‘that is to be regarded as en occupational expense required to maintain him es 2 Tro ductive resouree from final conswmption.* Pertaps the best procedure is to deduct, One of the ways of rationalizing the personal exemption and credit for dependents ‘under the income tax 19 a8 an allovance for occupational expenses of this kind. Similarly, the pressure for an "earned income oredit” derives from the recognition ‘that all expenses are deducted in computing taxable incone trom non-human capital, Bat not in computing tawable income from finan capital only those occupations) expenses that sre clearly special to a particular occupstion end to regard the mintnom expenses beyond thie necessary to atntetn the man being 28 a factor of production as the sone in all occupations. ‘This treatwent eccousts for ‘the initial seguent of sero net returns in the shove sketch, which 1s intenfed to afe- play the features of “typical” pattern of Life tine returns, The subsequent seguent of negetive returns refers to the period of training, when spectal outlays - for ‘tuition fees, books, equimsent, ete, = ere Likely to exceed any positive retumne. m™ereatter, in general net returns rise toe peak, and subsequently decline. Tn ad ition to the more obvious oscupationsl expenses, it 5 clear that income taxes should also be deducted in computing net returns (see section a below).“95 Bince the figures plotted are expected net returns, they conceal vide aiffer~ fences enmg the returns to different individuals ant are affected by the 1ikeldhood of unemployment, Sintierly, the declining seguent sn part reflects not only a possible decline in the productivity of the active yorker with age, but also the caller probability that a individual will be eotively earning ‘necne, the older his age, bectuse of voluntary rettrenent, retirement or idleness forced by 42 ‘health, or death. ote aleo that the vertical axis shove the returns 1f occupation A is chosen not from the practice of oscupetion A. Tt therefore includes earnings ‘rom other occupations that my be folloved instesd of A ty people uho choo a Anitially. the reason for this is that one factor affecting the attractiveness of assferent occupations 1e precisely the relative value that training for an oceupee ‘tion has in carrying on other ccewpations. White the shape of the curve drama shove 4s reasonably typical, 1t vill of course Gigfer trom occupation to ceeupetion in detail. ‘Te anoumt of capital tne vestment varies widely and vith 1t the age at witch expected eaminge Decoue poot~ ‘tives The peakedness of the curve anf the age at vhich the ponk 1s reached Leeviee very widely. Eipected net retame a fee ‘%e simple average level of lifetine earnings 48 not, of course, adequate to mumarize the attractiveness of « particular Lifetine earnings pattern, even to the Inpersonai slave-omer; in ¢ world in which the interest rate te net zero, the ‘timing of the returns matters also. For example, suppose the Lifetine earnings patterns for A and B ere a in the accompanying figure end that both have the same ‘ererage level. A is then clearly the more attractive finmetally, since the excess960 ‘earnings in Ain carly years could be invested at interest and so yield a sum not, favellable in B. To take account of this effect, ve can com pute the present capital value of the expected net returas in each occupation. Tet By B5y ---- be the expected anmval returns in years 1, 2, ---~, and x, the interest rate. ‘Then vem of Bf eee FF Fr? ‘is the capital velue in year zero of the strean of expected returns. TL will be recalled that the Life tine earning curves and so the capital veluee vere computed for parsicular wage rates. To sumarize the effect of the estegory of factors now under consideration, ve can ask what relative yage rete ‘would meke the capital values in the two ocewpetions equal. Suppose this vere @ wage rate Luh tines as high in A as in 3. We could then say that at this wage veto the tuo occupations would be equally attractive finsseielly or actuarially, end thet if actuarial ettrectiveness vere the only consideration, the supply curve Would be © horizontal Line at a relative wage rate of 1.4, as in the accompmying SESE HERE GS Figwe. ‘We nave elvendy Listed many of the Lah factors that will affect the relative vage rate thet will make two ocewations equally Roe attractive financially ~ geagonal and cyclical variability of emloynent, length of training, dizect cost of training, Airect occupational expenses subsequent to train~ ng, tex structure, length of working Life in an osewpation, tewporal pattern of curnings over the course of a working Lifetine. There are doubtless munerous others ‘that might be important for one or another partioular ocewpation, so @ complete statement te Smpossible ~ ¢ self-contained anslytionl apparatus for taking exch Factors into account is both yosaible and desirable; @ self-contained and complete List of the empirical factors to be taken into account 4e not.(b) Yartabiatty of returns ‘As already noted, the average net returns thet enter into the cepitel values Getined above conceal differences of return from individual to individual. ‘These sours hin dirterences are of Little importance to the slave-owner <+ at least if ve to om enough slaves == since they will tend to cancel out and so he cen concentrate on the expected return, fo the individual in our soctety choosing an occupation, ‘they cannot so easily be put to one side, te will want to know not only the present capital value of expected retume but also the datribution of retums = or more compactly, the probabilivy distribution of present capital values. Occupations A end B, for example, may be equally attractive financially, yet A may be an cecupe- ‘tion Like, say movie acting, offering @ snail chance of a very high revard togetter with e large chance of a small reverd, vaéle B may be an occupation like being a typlet, offering reasonable certainty of a particular retura with no grest chance of vide departures in efther direction. tthe effect of this variability depends, of course, on tastes of individuals ‘vith respect to risk or meertainty. If we sccept the expected utility theary of choice, the vage-rates thet will render two occupations equally attractive to en individual vill be those that will equate the expected uttlity from them rather than ‘the expected uoney return or capital value. If ait people hed the cae tastes with respect to uncertainty, the effect of Aitterent variability of retums would be to raise or lover the height of a supply curve Uke the horizontal one dram ebove for A and B at a zelative wage rate of 14h for A, If for example, A offered sali chance of a Inge return, alle B offered only moderate variability, and {¢ poople in genersl preferred the former Kind ot uncertainty to the latter, the effect of variability vould be to roiuce the height of the curve from 1.4 to a lover number, aay 1.3, the Atfference meesuring, ao it vere, the price people are villing to pay to get the kind of ‘certainty they like. For example, 1t {8 probably true that nost people prefer the Xind of variability aseribed ebove to movie acting to the kind ascribed to typing,“9 and tn consequence It 6 uy guess that average returns to morte actors = account veing taken of failures ae well as of successes ~ ie loss than average vetume to pists, Of course, people do not all have the sane tastes. Sue will prefer the ind of vartebiiity just attributed to A, some the kind attributed to B. ‘The fone cr will be attracted to A at a wage rate beloy 1d, the latter only at a higher vege rate, oo the supply curve vill be given a positive elope as in the accompany ing figure. If OA exceeds unity, £8 vould be reasonable to say that on balance people prefer the kind of variability offered by A, and conversely. rete inh Sage rete tb ah e 2 8 Wo.of man-hours in & (ec) Honpecuntery advantages In addition to the factors affecting the money returns fron different occupations, there are many cther factors that affect their attractiveness to any given individual = the kind of york tnvolved, the losstion st vhich 1t 4s carried on, the social prestige attached to 1t, snd so on. Like the variability of incose, sone such factors may be evaluated pretty mich alike by most people; in this case, their effect is to shift the ewply curve wwerds or domwerds. Insofer as people Gifter in their evaluation of non-pecuniary advantages and Aisndventages, the ef fect ts to impart a slope to the surly curve. Perhaps the extrene case of ditter fence 16 1f sone prefer occupation A over B end othere B over A, no matter vbst the relative pecuntary returns. In this case, the supply curve will be perfectly inelastic.oa Jf there vere no differences in tastes or abilities, and an essentially per- fect market, all supply curves vould be perfectly elastic and relative wege rates vould be determined completely by conditions of supply; conditions of desaad vould determine only the mmber in each ccswpation. In this case, all differences in re- ‘turn would be equalizing, and equalising to all individuala - that is, the structure of wage rates vould be such that each individual yould be indifferent viich occwpa~ ‘ion be pursued; there vould then be no “rente". At the other extreme, at which individuals are svayed exclusively ty nonpecuntary considerations, and there ere ‘vide ditterences in tastes, supply curves would be completely Snelastic and relative age rates yould be determined by conditions of demand. ALL wages would, as it were, be price-determined instead of price-determining and 60 be “rents”. 1a the more general case, in vMich there are differences in tastes and abilt- ties, but they do not completely determine the cholee of occupations, supply curves ILL be positively sloped. Tn thle case, aifferences in return vill be equalizing aly at the margin. Sone individuals will be receiving a rent in the sense that they would be willing to pursue their occupation at a lover total return, though even thes will be on the murgin in the sense thet they regard the addétional return from working © little longer or harder as just compensating for the additional costs in ‘volved in doing so. There is, that te, an extensive margin and an intensive one. ‘fm increase in denand vill tend to push the extensive margins outward by attracting nore individuals into the oceupetion. Its effect on the intenstve margin te less certain for the reasons discussed above in conection vith the backward bending short run supply curve of Labor in general. (@) me ettect of snecme taxes Tt sceus vorth singling out for special attention the effect of ncowe taxce, Hirst, because they have so greatly Increased in importance in recent years; second, Decause there {s such general misunderstanding of thetr role and auch widespread belief that they camot be “shifted”; third, beceuse they are omitted tron the List of fectors Atscussed explicitly by Friedman and Kuznets.-100- ‘As already noted, the relevant figure for the individual to compere in Judging the relative attractiveness of tvo occupations ts return after taxee, not return before taxes. It bas frequently been argued thet inccse taxes do not aftect this choice because the larger the incone before taxes, the larger the in cone after taxes, and hence Af one occupation is more attractive than the other be- tore taxes, it vill also be after taxes. Unfortunately, thie 1s not trus, partly vecause the base of the tax 1s not the sane as the figure that 1s relevent in con sidering net pecuniary return, partly because the tax dase cannot take account of non-pecuniary factors. Consider first @ straight proportional incone tax with no exeuptions. Bren ‘his tax will affect the relative retums in different occupations. The uost obri- ous reason will be if the tax base does not permit the deduction of all expenses regarded as occupational expenses in choosing between oceupations and these differ from occupation to cceupation. fit if the tax base is the sane in thie sense as ‘the return relevant to the choice song occupations, it 1s almost sure to differ in other senses. For example, let one occupation yield a return that varies fron year to year for any given tndsvidusl and 1s cometines negative whereas another yields ‘he sane incense during each year of vork. Unless the tex provides for @ subsidy (a negative payment) when net income ts negative, the tex burden vill be hearier on the first occupation then on the second, 20 that at a relative wage rate thet would ‘uske the present capital value of the tvo occupations equal before taxes, the capi- tel value vill be enaller in the first occupation after aLlovance for taxeo. The pertieular effect 1s ty no means a curiosity - it arises especially as betveen cocupations requiring training and those that do not, since in the former the re turn 48, as we sav earlier, Likely to be negative during earlier years. In these cases, the neglect of negative incones 1s the seme as not permitting the expenses of training to be dedueted 1m computing taxable income. ‘the effects so far considered could in principle be eliminated by proper aefinition of the base. But thie ie hardly posetble 1f tvo occupations aitrer in-101- non-pecuniary attractiveness, so thst @ higher xonetary return 1s requtred in one then in the other to make then equally attractive. Ih this ease, 1t would take @ igor relative return with the tax than without {t to make the tuo occupations equally attractive. In effect, the nompocuniery advantages of the lover-pald occu pation ia not subject to the tax, oo that one vay to avold the tax 1s to engage in occupations vith large norpecutary advantages. Tue introduction of an exeption and of progressive rates hes additional effects. am occupation in vhich an individual's income fluctuates from year to your vill tend to be more heavily taxed for a given present value before tax than one in waieh it is constant from year to year. Here again, changes in the tax lav, to provide “averaging” of incone might eliminate this effect. Dut no changes can very well eliminate a comparable effect when the variability is betwe people. Suppose An advance occupation A and B promise the sane average incone before tax, but fu A Sncone verdes nore from individual to individual than dn B Then with @ progressive tax schedule, average incone after tax vill be lover in A than in B. Tue progressive tax accentuates the effect via nonpecuniary advantages mentioned stove. For vith such a tax, the ratio of incomes after tax vill be lover than vetore tax. Ye follows that the extotence of an income tax does afvect the choice of ‘occupations and so the allocation of resources among different uses. Indeed, if all aitterences in income yere equalizing, in the sense that supply curves of the kind ‘we have been draving vere horizontal, an incone tax yould have no redlatributive effects at ali, no matter hov progressive. Relative vages after tax would be the sere with a highly progreseive tax as vith a flat tax. The reason 1s of course that people would leave cccupetions epeclatiy affected by the highly progressive tax (occupa: ‘tions that are Bighly pata to copensate for extrexe nonpecuniary @tsadvanteges, or ‘hat offer highly varieble returns, ete.) and enter those less affected by st until ‘hia pattern of relative vage rates vas attained. The same relative wage rates after tax vould of course moan higher vage rates before tax in the occupations102 affected by Uhe Mghly progressive tax and this vould curtatl the quantity demanded to match the reduced quantity supplied. Mare generally, of couros, differences in taste vill produce a positively sloping supply curve, eo that the form of tax will effect the relative vage rates. ‘The reduction in numbers employed in the occupations specially affected by the highly progressive tax would be produced by the exodus (or better fatiure to enter) of those who had the least attactaent to these occupations on non-pecuniary gromas. ‘The final result would be @ lover relative wage after tax than vith a flat tax, though, of course, a higher relative wage before tax. Th ts clear thet thie anslysie of the Income tax paral the ueuel analysis of excise taxes, And indeed, 1t seems likely that corresponding to any given income tex, there ie, in principle, sone set of excise taxes on final services that vould have precisely the cme allocative ant distributive ettects 2. Digferences arising fro noncompeting arows. order for differences in return to be prodeminatly equalizing - that ts, produced by the factors iscussed in the preceding section = it ts clearly nece essary that many individuals be in a position to choose freely between the occu ations in question. Wow to 8 very large extent this 1s the case and, accordingly, many exteting &ifferences in wage rates can be regarded equslizing differences. But there ie considerable evidence that not ell differences in return can be 50 regarded. Tn partiouler, dizgerenees in return between euch trond classes of occupations as professional and nomprof sional seem considerably larger than ean bbe explained in terms of differences in costs, nonpecuntary advantages or dis- aavanteges, and the Like. ‘Te additional factor that entere in in such cases te a barrier of sone kind or other to entry snto the better paid occupations. Only some individuals are free ‘to choose these occupations; they constitute, in Catmes' happy term, a “aoncowpeting group". Many Aifterent causes may give rise to barriers to extry and so to the establishment of noncompeting groups, and it aay be desirable to List some of the7103+ nore important. (1) Deliberate restrictions on entry. Imnigration restrictions, for ‘example, make fnerican vorkers « noncompeting group relative to vorkers in other countries. Within the country, the requirenent of a icense to practice an occupation - as in medicine, lav, and the like - may be a means of deliberately restricting entry. ‘The granting of Licenses is generally placed in the hands of people currently in the oscupation and they have an understandable incentive to restrict entry. Again, trade union pover to foree an employer to pay no lest than an agreed vage is @ means of restricting entry into the occupation. Festrictions of this kind are extrenely muserous in detail. But, hovever ‘vexatious, T would judge that, except perhaps for the immigration restrictions they have not been of major empirical importance; almost surely not important as some of the other barriers to be mentioned. (2) Geographic Imobility. This is often ofted tn return, particularly for alleged differences between North and South and cause of aisterences country and city. It seems doubtful, hovever, that except for particular end Aeolated cases, 1t can be of any major importance in the United States. Census figures show quite extraordinary movenents of people - during the 1940's, for example, the roveaents vithin the United States quite dvarfed in mgnitude the forced movements of population in Purope = both these forced by the Nazis and ‘those forced by the Soviets. And it mst be recalled that it is not necessary for everyone to nove ~ mobility at the margin ss enough. (3) Differences in ebilty. Tb ie aoseviat arbitrary whether to reward ateterences sn ability as cresting noncompeting groups or to conbine then with @ieterences in taste and regard then as giving rise to equalizing differences. Te As clear that they will produce ifferences in returns as between individuals greater than ie required to compensate for Aifferences n conte incurred and the site j An effect, one individual 19 more units of Isbor pover then another, more Imunen capital. The effect on vage rates in identifiable occupations arieesom decmuse éifterent occupations will tend to be staffed - or to require - different average levels of ability. Of course, there 1s no objective standard of "higher" and "lover" abilities that will be respected by the market: whether particular ‘type of ability vill be highly remunerated depents entirely on whether the densa for t 4s high zelative to the supply available. Sone examples ney perhaps shoy why it is difficult to distinguish aifferences 4m "ability", in the economeally relevant sense, from differences in "taste", and why 1 Le tempting to include then with the factore giving rise to equalizing ditterences. Is the relatively high compensation of a deep-sea dtver to be re garded ae a revard for the sonree ab{lity of being willing to york under yater and in dangerous circumstances or for the nonpewniary Gteadvantages of the trade? Woot of the stunt-artiet? the phyeietan? Obviously there is a Jarge aren where “ability” and "tastes" merge. (4) Socto-sconomie etratitiontion of the scetety. In many countries tt 48 still true and in most countries 1t was true not eo long ago that perbaps the uajor source of internal barriers to entry was posed by the stratification of the population into soctal classes. In general, the learned professions and certain other occupations have been freely open only to mesbers of the upper classes, and eo on dovn the Line. Of course the stratification Je never complete, there 4e aivays sone possibility of upyerd mobility, tut the hindrances in the path of such mobility suffice to keep wide difrerentiale in rate of return. hie kind of strictly soctal atratification hes never been ae important in ‘his country us in most others and it has clearly been decreasing greatly over tine, in large measure because of the wide availability of education. This is clesrly revealed in the behavior of relative wages in clerical and manus) professions Literacy was at one time sufficiently rare to give rive to a noncompeting groups At clearly is no longer. In consequence, there has been longstern downward ‘trend in the ratio of earnings In clerical pursuite to earnings n mnual pursuits Yrom being conciderably higher paid, clerical pureuite are probably noy im general,-105- lover peld. On a higher level, the same phenomenon is repesting itself in the ratio of the salary of college teachers to the salary of high school teachers: this ratio has been declining steadily over tine. ‘The difficulty or impossibility of having a good capital market for invests ent in hunan cepitel 1a 2 uaJor reason viy social and economic position cen affect ‘the alternatives open to @ young wan {n choosing his career. ‘he possibility of getting exensive training depends on the eb{lity of » parent or Denefactor to Finance Lt, or the viblingxess end capacity of the yousg man to “work his wey through", and even then, on the ab{lity of the young man's fondly to do without ‘the earnings be might othervise get during Me training. These factors reaain important for certain careers and doubtless are one of the most important source of differences in earnings attributable to noncompeting groupe. (5) color might have been included under the preceding heading but it sens better to separate 1 out for special treatment. Clearly, negroes have not been in the cane poattion as whites to choose among occupations. They have not had the sane possibility of getting training end education, partly because of lessened availability of public facilities, partly because of Aiserimination in private institutions. But the effect of color te mich more complicated than thie, Be cause of prejudices of both custoners and fellow vorkers, being @ negro involves having @ lover eeononic productivity in some occupations and so makes color have ‘the sane effect on oarmings a5 a difference in ability. As a result, the stretifi- ation of the populstion by color has clearly been one of the most potent forces producing non-oqualizing differences in return sn the United States. 3+ Transitional gttferences in return ate heeding requires very Little discussion. Clearly the eupply of stor of a particular Kind te Likely to be much less elastic in the stort run ‘than in the Jong rin, 90 any change tn deuaod 4¢ Likely to have mich sharper eftects initially than ultinately. Perhaps the only point that needs fuller siustration is the potnt made at the outset that viat we eal @ tranet tional