Implicit Contracts and Underemployment Equilibria Costas Azariadis ‘The Journal of Political Economy, Volume 83, Issue 6 (Dec., 1975), 1183-1202. ‘Your use of the ISTOR database indicates your acceptance of JSTOR’s Tecms and Conditions of Use. A copy of ISTOR's Terms and Conditions of Use is available at hitp:/www jstor.org/aboutrerms.himl, by contacting ISTOR at jstor-info@umich edu, or by calling ISTOR at (888)388-3574, (734)998-9101 ar (PAX) (734)998-9113. No part ‘of a JSTOR transmission may be copied, dawnloaded, stored, further transmitted, ransferted, distributed, ateced, or otherwise used, in any fotm ot by any means, except: (1) ane stored electronic and one paper copy of any article solely for your petsanal, non-commercial use, ot (2) with prior written permission of ISTOR and the publisher af the article or other text Bach copy of any part ofa JSTOR transmission must contain the samie copyright notice that appears on the seteen or printed page of such transmission, The Journal of Political Economy is published by University of Chicago. Please contact the publisher for further poimissions regarding the use of this work. Publisher contact information may be-obtained at lttpsferwwstor org/jounals/ucpress html. The Jounal of Political Economy ©1975 University of Chicago ISTOR and the JSTOR logo are wademarks of JSTOR, and are Registered in the U.S, Patent and Trademark Orie. For mare information on ISTOR contact jstor-info@umich edu, (©1998 JSTOR baipfsvwojstor.oug/ ‘Sun Aug 23 00:33:40 EDT 1993Implicit Contracts and Underemployment Equilibri: Costas Azariadis rote river his paper studies aa industey wich demand uncertainty which prompts riskeneutral firms 1 act bath as employers and as insurers of eoine: acneous, visk-averse Iaborers, ‘The revalting conitaelual anangeneats ure: ous in their simplest fare, be mare likely 12 specify fall employ» tnient the more of the (ellamine conditions prevail: small variability iu product peice, above-average ecenomy-wide labar demand, highly risk averse workers, small unereplaymient compensation, and highly com petisive product markes, Ocherwise, it may be optimal fer fens we lay BIT, by eandors choice, past of the work fares during low staces af demand L Introduction Competitive vs ie theney predicts that firms will adjust «@ a contraction, in product demand by lowering both employment and the ceal wage rate, in contradiction to the normal industrial practice of laying off unneeded workers and paying unchanged wages to the rest of the wark force. This parados,* which arises frorn the close relationship betwee competitive equilibria and Pareto optitna in auction markets, spawned (his paner based os chapter 2 a's dectaral dingoracion subensted the Seed ‘of Jadyeeil Adluustraion, Carnegie-Mellon Universe, January [976 Easier versions ‘wore peesenied a the Riscnran toto! the enmaneetite Seciery, Oss, August U7 and au the Lisivenity «f Chicaga Money ane Basking Worlshep, Decewber (178 Tam indebted 09 Robert b kuene and Eésvaed Present for innumerable digeusinn to Rateet Barr, Havil Cae, Herschel Gress, Jahn Kerman, Castaphee Sit, 2 Xb aavonyavuna refer ow eansructive eich; and t@ the U.S. Depacrenene of Vakar Joe nancial sniott. All ate abeolved fears ay si gf oatision oF evasion "Ty quot fro Areas. 147%): "One of he vessteriaus dhings is wh he empayers] {0 cpt eu wages Le ny be that they wally eo ot as isha is goign tera the fpoeble need te eeu Inline inthe fotuer sind ate wateiod aba the impact of any sae Ets on their alley te prec om ee Tabor mare later.”edly JOURKAL OF POLITICAL RooNOMY substantial corrective effort by economic theorists, One reaction, exernpli~ od in che weitings of Keynes (1936) and of 2 diverse group of epigones (Posinkin 1956; Clower 1966; Leljonhutvad 1958; Barro and Grossman 1971), has beet. to abandon short-run market clearing as 2 workable \ssumprion and focus instead on quantity adjustments as the primary ieans of response (a market disturbances. A far less systematic, but intuicively more transparent, commentary? has appeared periodically in the hterature, arguing in effect chat devi- asians from Parcio optimality occur because continuous-auction models cannot do justice 1 certain attributes peculiar to the labor resource. Chie among these actributes are the limited mobility of the labor re- sowcce andl the difficulty of diversifeing human capital under uncertaincy The relative immobility oflabor over markets, especially those involving dliflexent industzies or geagraphical locations, is fairly well documented.” However, the obvious fact bears repeating shat no markers exise for che dlinvet exchange of claims on future Labor services;* the costs of monitoring, andl enforceraent, and “mora hazard," are some reasons why such mare ees have not arisen. Nevertheless, lease part of che risk an uncertain labor income stecam eoates for its recipient can be shilked 19 third parties by employee inter mediation. tat is, by the tacit or open conuitmnent of tke firm co guaran: tee its personne! that divir wage rates, hours worked, employment status, or combination af all such frctors, wit] be in some degree independent sf the vieissitudes of the business evele.® The risk is thereby eransferred, Tran wages 19 profits and, via the capital market, the income streams oo the firm's oveners and creditors This process is subject to (wo limitations. First, assurances of the sort just desevibed will not be handed out evenly to all personnel. In breadth snd fieiaacss of commitment, employers will discriminate in (vor of ppeesons in whose sesating substantial investment has been oF is about to fre lade. Others whe are ehought to possess superior qualities of produc. tivity, reliability, adapeability, ete., will be similarly favored. Second, a.xulional enteepraticur will not shield dhe terres ofemployment af even his ‘lost valualile group of employees against arbitrary demand (hsctuations Ul daing sa invalves more then a token probability thar the fem might go bamkeupt Neither qualification should be a kindrance in che short-run, homo- 2 bow interesting eaelyspeciouens, ee Knight (1921, chap. 9) abd Hick (293%, chap. J). SL the weidernen Taree and Bugcon (1971) * Criss could eqiesivably sores hat fonetion sere it wot geceeally che ease that incynicy a the members of any sie union are highly correlacee “The digcussion chases Ressily og Brainard and Dolbrar (1971); se alse Tobia (19721 vad Gordon (L970. ‘feigtn this eran that Knight (L921, yp. 271-72) characterizes wages an “contractual” ines ane rate a “ers” incomeUNDEREMPLOYMENT EQUILIBRIA 1185 geneous labor problem that is studied in this paper; itis sufficiene that, in the eourse ofthe relationship between employer and employees, enough scope exists for the former to uniourden the latter of at feast same of the variability that otherwise would accompany wage income. At first approximation, one need not worry whether risk shilting oceurs merely ecause the firra’s net worth far exceeds thar of its average employee or because it is in the nature of hiring that “the confident and venture- some ‘assume the risk’ or insure’ the doubtfui and timid by guaranteeing to the latter a specified income in return for an assignment of the actual. results" (Knight 1921, pp. 269-70} “The drift of the preceding arguments points to a more complex view ofthe labor market than is customary in conventional short-run analyses: in uncertainty, labor services are not auctioned off in quite the same way fresh fruit is. Rather, they are exchanged for some implicit set of corazait- ments, hereinafter called an. impli dabar contract, on the part of the fren fo employ the owner of those labor services for a “reasonable” period of time and on terins mutualty agreed upon in advance. Tn such a marker, job choice will depend on what subjective value prospective employees place on alzernative contracts, and nat on the value of any one contract component, After all available contracts have been compared, and before the state of demand in any industry is known, suppliers of labor services will each gravitate toward the highest attain- able value. The initial distribution of the labor force thus achieved will be independent of the state of nature: it will also tend to have some potmianence, as indicated by the large proportion of rehires in inanufsc~ uring accessions Teonracts, then, are not exsily abrogated, employers should possess considerable discretion over specific terms of employment. The possibility arises, for instance, that a jab which requires the same teclnicai skills, as anather will pay a different wage. The freedom to set specific compo= nents of contracts at a level above or below what prevails elsewhere is limited by the familiar property of any equilibrium distibution of the. labor force over firms: no ceonomie agent values the bargain he/she has actually agreed to less than any other feasible contract.” A systematic study of these issues begins in Seetion IT with a stochastic, environment in which risk-neutral entrepreneurs and identical, risk- averse workers with indivisibte leisure endowment exchange a particularly simple type of contract. Section ITI examines the conditions under which, the wage schedules of labor contracts will be invariant to changes in relative product price, A necessary and sufficient condition for the sub- optimality of full-employment contracts is developed and interpreted in Section IV. A complete characterization of the optimal contract fellows 9 Cf, Stiglite (0974).11365 JOURNAL OF POLTICAL ReONaKY in Section V along with 2 comparison of its provisions with what would prevail ia a labor acution marker beset by identical stochastic distur- dances. A summary and conclusions appear in Section VI. UL Uniform Contracts Suppose that the quantity of a perfectly homogeneous, nonstorable commodity demanded of an industey depends on the prevailing price, p, of the commodity relative to that ofall other goods (the fixed, exogenous “price level”), on the state of nature, 5, and, possibly, on a vector of nonrandom, nonprice variables aver which the industry fas no contzol. For simplicity, assure chat the actual state is drawn randomaly from set of discrete states S = [sls = 5, s3.--., 3) according to a known probabilicy distribution 4(s}, for which Dae = w S ‘The industry consists of a fixed number of identical, perfectly compet- itive, risk-neutral firms to which changes in sare revealed through shifts in the product price they can charge their customers. Let pls) be the (pacamctrically given) mapping of states into prices and, without loss of generality, suppose that, for any , 53 in S, HS 52 A5;) < pla), (2) ‘On the other side of the labor marker there is an even larger number, M, of tisk-averse workers wha are identical® in tastes, endowments, and technical ability and who may or may not have alternatives outside this industry. Assume, for simplicity, that these workers command one indivisible unit of leisure and that their preferences over consumption, « and leisure, J, are embodied in a monotone, bounded utility function ale, tj, concave in 6? Let the random variable y denote current labor income; the constants a and ¢, denote, respectively, curvent property incorne and the maxiraum, level of consumption attainable when y= and ! = J. That level may exceed « by the amount, if any, of unemployment compensation which accrues to the worker. Without loss of generality, we set alo, 0) = O and weg, I) = K (3 "This assumpzion serves ta keep down the disiensionality of the contrac: space. For instance, if seotkers differed ystematically ia prclerencn oe in their endowment af human or nonhuman wealth, then conteacts indeed ov all of these eharacteraties wale generally be appropriate * Observed labor reference foc lumpy leisuee sears (long weekends, paid vacations, fe] taise a strong stspicion that oie mot concave ix {for Bigh enough vaiies af leisure conaurapticn, Th? implications for unemployment af peelergneas locally nanconex ik Ieinuee are examined ig See TV.LUNDEREMPLOVMENT EQUmLEARIA 1187 and dofine an increasing, concave function u(+) feom u(y] = of + 9, 0). (4) ‘The range of contracts offered by each firm will depend partly on how varied its work force is ia skill, attitude toward risk, and the like, and partly ‘on the lengtli and the administrative costs of striking, monitoring, and enforcing cach agreement. The most important nonwage component of such agroements js probably “job security,” which depends crucially on. variations in employment Suppose, then, for simplicity, that all freely agreed upon coneracts are strictly enforceable, perhaps because “cheaters” suffer a catastrophic. lass in reputation; are of the sane, institutionally fixed duration; and pay no direct compensation to unemployed workers, Since. workers are, by assumption, absolutely identical, a uniform contract will likely be affered to alll of them. This will be 2 random vector of the form 8 = (ula), als), oy which reveals whar wage relative to the price level) and volume of ‘employrncnt the typical firm plans to offer in each state. Valuation and Demenance ‘The valuation af a contractual offer by laborers is clearly contingent a the ultimate size, m, of he firm’s work force. The larger the number of people who consent to a given contractual offer before the state af nature iy knowa, the smaller die (equal for all) probability, p(s), that any af them will end up employed in state +. Tn fet, ats) Let ¥(6, m) be thee expected utility from a contract 8 which attracts m ‘workers, thurs the expectation in F should be taken wich respect to both ‘and cmployssent status, Regall that the utility of working in sis afa(s)) and of not working is K. Hence, mie [1 m(siid (a V@,m) = Blue) + - pAb (Sb) From {a} and (35) it follows directly, as one might expect of a compet- itive lalinr market with mabile Laborers af known quality, that all vacan- lex will be filled. Hence, els) = alsin 6) ‘0-Thin in a atateenent about averages. Ik docs got mean ta deny that che frm could [eiGitaldy ue nines than the eneiracted amount of Labor services in 4 pactislarly favor Able sate of nacure ut, eather, sae ie eannar de so.on the bass of information available A thr oe cimeaete ae seve1188 JOURKAL OF POLITICAL EcoNOMY ‘The actual value of m corresponding to any contract offer will depend, naturally, on market alternatives. Let 2(K < 7) be the market valuation attached to such offers under an equilibrium distribution of the labor fore, Then the right-hand side of (3b) rnst equal A, and mis a fonction of both and the components 2» and # of the contract Some usefinl nomenclature follows, Tn definition 1, a comteact § = {wo(5}, n(s)} is called (i) faaribie if (6, m) = 7 for some positive m such that ns) & m for all s; (ii) full emplosment (underemployraent) if n(s) = for all s (< m for some s); and {lit) fixed wage (variable wage) ifm is (is not) independent of s. An example of an infeasible contract is n(s) > 0 for some s, Bulwv) < K. A Paretian criterion for comparing contracts follows. Defaitin 2: Let m be the size of the work force attracted and x(5,) be the expected profit accruing to a firm under a feasible conteact, 4,. Then 6, is said to (f) daminate 5 (8, & da) Hf 5, is feasible for m and alb,) & n{8}; and (ii) be opfamat if it dominates all feasible contracts. ‘These definitions suggest that the optimum contract, 5*, is simply the solution to a well-defined constrained maxinwum problem. The primary im of this paper, however, is not to characterize 3* fully {which is done in Sec. V}) hut, rather, 9 find out whether 8 is a full-employment con- tract. In Section TTI T reduce the dimensionality of the fem’s opportunity set hy excluding contracts that are demonstrably infertor. TIT, Wage Rigidity Two sources of variability in wage income are present in the class of contracts set farth in (5): one arises from the stochastic nature of the wage schedule, the other fram uncertainty over employment status. That under certain circurnstanees itis profitable (far bath sides) to remove the foriner is shown in Lara 1: Given any feasible variable-wage contract 3, = {ee(s}, x(s]}s there cxists a feasible Sxed-wage contract 5, that doininates 3, wich strict dominance if w(+) is strictly concave. Proof: Let m satisty V(B,, m) = 2, and define 3, = (4, nls]} where i = E(unjJEx. The contract 54 produces the same expected. labor income far workers and the same expected cost and expected revenue for firms as does 6,. Note also that As VES ym) — Vm) = Ex{utid) ~ w)] > (>) lined — 1) u'(), by concavity (strict concavity), WR) WwEn — Eun = 0 Since V(85, m) = A, there exists a nonnegative (positive if w is strictly concave) constant « such that the contract 6, = {i — 8, nis)} satisfies Vidq m) = Fy m) and x(8,) 2 2(3,), QED. 1 Fora setae rule in an interermpors} context see Bally (1974).UNDERENPLOYMENT EQUILIBRIA 189, A direct corollary of this result is that, in an expected-value sense, enforceable labor conteacts dominate labor auctions. Let ze(r) be the {parameitie) wage schedule associated with an auetion market, /{-] be the production fiction of the typical firm, and F(-} be the inverse of the marginal product function. The outcome of the auction is then equivalent t@ the contract J, = f(s), Flals)iie]}, which, for some nonnegative <, is clearly dominated by the cantract 6 = (Eia(s) — 8, Pho(sifatsily ‘Too further points should be emphasized. First, workers who are neutral toward risk are indifferent between a variable-wage contract 5, = {w(s), n{s)} and a fixed-wage one 5; = {Bufs), n(5}}. Second, the primary use of lemma f is to reduce the complexity of the ensuing math= eratics without affecting the ultimate results, A literal interpretation ‘of it would he misplaced, for it is not rabust to changes in assumptions about the indi y af leisure, the enforceability of contractual bargains, the nature of the randae disturbance, and, possibly, the risk ‘af ruin to the fires. With variable hours of work, for instance, wage ine come a{1 — 0) becames state invariant if preferences ate additives xf nol, a more complicated function of income and leisure will still be non stochastic Nor should one jump to the conclusion”? chat the real wage schedule is invariant « random disturbances in aggregate demand, In fact, theories of speculative labor supply! seem to argue against such invariance: when prices are higher than average, workers desire to accumulate assets faster than they orally do. ‘To support a higher flow of savings, they will increase their “propensity” to save and accept a somewhat lower wage in terms of commodities, so that commadity producers have some incentive to hice a largerthanenortnal amount of labor services In addition, few real-world agraexaents can prevent workers laid off in a temporarily depressed industry from auctioning off theie services to the highest bidder in another industry. What effect these instances of irmperfeeily enfarceable contracts will have on the wage schedule industey is nat immediately abvious, but, as a brief argument in Section IV suge goats, the outcome will likely be a softening of wage rigidity IV. Ave Full-Employment Contracts Optimal? Let ® = € 4, ) be the class of all full-employment contracts that are feasible for the typical fem, ane imagine that we isolated the dominant meraber, 5¢ = (uf, mt. of that class. Consider now the class D = ¢6 > 2 One might be éemptcd in that direction by the apparent (eilure of econoracirie wack tc find a sigeuicont carclation between tend-cartected average eal compensation per man-haur (obviously a pracyclically based proxy for the eeol wage tate) and any peice index. Cl-Bodlin (1969), wha alka ceviews earlier efforts im hat fel 1 See Tobin (947) aud Lucas and Rapaing (1969).trae JOURNAL OF POLITICAL ECONOMY of all feasible contracts with ti same labor farce as 6 which ate formed from it by reducing emplayment below min ac least one states that is, 3 = (w, n(s)} where w is a nonstochastie parameter at least equal to wf and nis) < mf for all s, < mi for some s 6) ‘To continue the process of whiteling away at the firm's opportunity: set, we ask now whether there is an underemployment contract in D which daminates 47, Note that full-emplayment agreemctits possess two desirable features, namely, zero variance in Jabor income and the lowest wage rate consistent with atomistic competition in the labor market. Against these advantages one must weigh the drawhack af a stave invariant employment schedule which does not permit fill-cmployinent contracts to vary work force direcily with the relative product price, that is, to use a larger amount of labor services in more profitable states of nature, Every underemployment bargain in class D will pay 2 wage w > wf (© compensate employees for the risk of temporary job loss. For small risks, the ausrage undevemplaymant premium, w — uit, can de computed from the requirement that all members of D are feasible contracts. Thus, if ia Es, A= Gilt) ulao) + CE — fmt) K => eA — RY = sfuten) — KY Note that (wf) = 2. Expanding uu) about wit and neglecting terms of order higher than first, we obtain mJ — K) =A + (w= wf) wwf) — KD, (6a) whence (er = why = mf — 8) A — Ki te a The expected wage bill, wit, ofany 4 in D may be thought of as consis King of ewo parts: a divect expected cost, 2978, equal to che minimum dalla outlay required wo secure the average amount of labor used; and a premium expected cost (w — wf)k. Expected profit from contract 8 is then #8) = Blofsia(er] ~ mf nbs) — Io — a = K)la'twH). (8) Now tet Plmj) = (2 — Kutt) = (uh) — Kueh) o denote the marginal underemployment premium, that is, the increment to premium cost required to compensate the labor force fot the loss of ‘one jab in any state. The profit contribution of the mth taborer is state ris Bs) fe) ah + bleeh), cayUNDFREMPLOVMENT EQUILIBRIA tage * : ‘ ‘ \ wp bw) peed ‘ t plsj)f' tn) bo, ™E Fig. 1.—Inferiosity of full-emplayraent contracts If there is a siate # such chat MELLO) + wh) < wh, oy it follows chat, relative to 49, expected profit will be raised if workers are laid off in state J. For fulbempleyment contracts to be suboptirat, it suffices, then, hat there exist a state of nature in which the fully played wage rae exceeds the sum of the marginal product value of fully emplayed labor plus the marginal deren ployment premium, Tae implication is (fig. 1] hat, if 5, is che worse stats, fial-employment contracts will be suboptimal. Ic is not clear what type of conuraet is best when s, is the lowest statc—ualess we are certai thae (14) is both necessary and sulficient for the inferiority of full-employ- ment agrecinents. ‘This is examined next Phe Macmum Problem The fixed-wage feasible conteact thas firms offer will be che one that generates the highest expected profi, Given any 6 = fw, a(s)} wich ufw) > K and 4 positive number m such that uopiue} + = nn] K] eo me = = RY "Le = RYEn, feasibiliey is equivalent to um, oF MG] < (A — KY *(atea) — APE gs} ntsy (12) Expected prafit is (5) = Dabs) (Alsi f(n()} - wnls)}, (13) "might be, far instance that chen 4 ie the worse state 9F is dorninated by an underemployment eaztract which dacs nat helng to elon Dtrge JOURNAL OF POLITICAL ECONOMY. where the production process wes one single input, labor, and satisfies the usual monotonicity and striet concavity requirements, We assume, farther, char F(R) 9 Basn +9, fH) + asus oo. a4 The decision problem is then to find a nontrivial solution [ie., such that nfs) > O for some 5] ta salve the probe (P] max 2(3} ‘al subject to (12) and nonnegativicy, Using the one-to-one mapping (en) > (nye = alee) ~ A), one easily shows Lemme 2: (P) has at least ane nontrivial solution, exzetly ane if wis strictly Let p(s) be te set of nonnegative multipliers corvesponcling to the vet of inequalities in (12). 1Cfollows from lemma 2 and (14) that the unique solution to (P) i completely characterized by the (KulvneLucker} conditions A + ute) Dale] = 6) = 0, a9) aN (ODF nak — aw + KY Mele) — APE eC} = ee) YH (18) KY "Lafee) — KIS a6) #66), = itl) > Vs 17) Equation (151 confirms lemma 1 and is useful in eliminating T pls] from (161. Phen nu) <= alef@n ~ 2u)) =n Ys (18a) where the wage rate net of the marginal underemployment premium, she) Sw — [u(w) — K] Judo, gy represents the ceductio job in any state in labor cost accomplished by eliminating one The Leferiorty of Full-Eemploymient Coniracts: A Necessary and Sufficient Condition Let # be the inverse function u~' From feasibility and equation (15), one readily abtains Lemma 3:30 = IANS mp FURQUEP] We can now formalize the preceding argument in Theorem 1: ‘There exists am underemployment contract, 5*, whick dominates 57 if and only if PULL) < zh) for some se & ag)UNDEREMPLOYMENT EQUILIBRIA 1193 Proof: Sufficiency. Recall that (3) 2 0, ¥s. IF (19) holds, then 8 clearly violates (J6a) for some state; neither it nor any other full-eraployment contract can then be optimal. Necessity. We will use & constructive proof of the contrapositive statement, that is, that BASF) & awh) Yee S 20) implies optimality of 3}. Indeed, if (20) holds, then we can define nonnegative muleipliers 4s) such that Bre] = Oot mt) — xe] an) Summing bath sides of (21) over 5, one obtains Eure) = PORE — eleaf) = (2 — Ki eee) by (18) and lemma 3, [¢ follows that the vector [w, m¥, u*(s)] satisfies conditions (15), (/62), and {17} and, therefore, (15), (15), and (17). Tf (20) is crue, then 87 is the optimal contract, QED. This yesult is helpful in reducing the issue of underemployment ever being optimal, in an expected valle sense, to one of whether of noc a given inequality holds. What is iin labor force preferences of consump- tian versus leisure, in their attitude toward risk, and in material endow- ments and industry demand characteristics that makes (19) tnore likely, or less likely, to prevail? Itis useful to limit temporarily the range of attitudes toward risk under consideration by assuming that the indes of relative risk aversion, Ps Ate + whe (wh i'w), (22) isindependent ofa + w,' We can hus concentrate solely on preferences ‘embodied in linear transformations of the following funetions:'* uw) = (a + ifr< (23a) = log (4 + we) fra (238) = Heme fro (23¢) L summarize in the next two lemmas some usetul bits of information, Both proofs are straightforward Lemma 4: Let f= '\X) be the wage rate which makes workers in- different bewween work and leisure. Then the function 2(w!) is decreasing 5 Recall that uf) = ale + am 0), where the constant a is nanlabor income. era (1971, pp. 82, 163-4) argues that'r a weakly ineveasing in wealth parily oa the basis ff empitieal ndings that the wealch clayicicy of money derand is at least | (which reales money a “luxury” gaod). The evidence, haweeer, seems less than averwhekming Shot demand far eash balences hat signin highee than unitary clascity (Meltzer 1963, pp. 736-38; Laidler (368, pp, 96-102) “eSee Pract (1904, pp. (95-35) for dette9g JOURNAL OF POLITICAL, ECONOMY for all w > &, is invariant to linear transformations of the utility function », and satisfies 2(f) =k. ‘The lemma immediately below is based on the assumption that ris constant. Lemma 5: The equation z(w} = 0 has a unique root, tia, which satisfies G) wy = ke = a ifr = 1, = OO — gifs #1, where e is the natural logarithin base; and (ii) wy is a decreasing function af r such that wg + 00 as + 0, + —aas7 -+ 0 ‘These lemmas point out two factors, tie workers’ opportunities outside the industry and their attitude toward risk, which bear on the optimality of full-employment contracts in an industry. For a large enough value ofthe index of relative risk aversion, ig will inevitably fall short of wif and hence 2(w{] < 0, which rules out (19). Underemployment is more likely to occur, om the other hand, for values of r low enough to place (tf) in the interval (0, £j, Similarly, the higher aggregate demand is, (or whatever other exogenous parameter happens to define the economy wide outlonke for these workers), the more attractive working is relative to not working and the larger 29} becomes relative to aig. This lowers the probability thar (19) will hold.!7 Other determinants of the likelikood of layoffs are readily apparent in (19). Layoffs should be more frequent in industries with relatively volatile (high variance in s) or relatively inelastic (p highly responsive ta) demand schedules, in markets with relatively Little competition, and in cases of relatively high unemployment compensation (which boosts K and thereby depresses the marginal underemployment premnium).)# Divisible Leisure If one were interested in what statements a theory of labor contracts might make about, say, the behavior of hours worked over the cycle or about seniariey claims on overtime, the obvious extension of the simple contract form is 8 (wls, 0, 266.8), 26, Ob, where 2(¢, t) = 1 = Us, ¢) is the amount of hours worked by each laborer iff state 5 occurs in period ¢ As one might expect from Section LIL, such "7 Since te constant Fin lemmas # and 5 depends on its wot abvious i this raadel how the size of nonhuman wealth affects unemplayment, Far a fisd economy-wiid= filemployment wage rate, both working and not working become more attractive 28 rnonlabor income rises fan the ether hand, wealthier workers were able ta hold out a they might in a search-oriented model—~tor ewer opportunities than could thelr poorer colleagues, nonhuman wealth and the probability af accepting a uniform. contract ‘offer right be negatively carzelated (see Danforth 1974) °" Hor additional facts afecting layoffs, see Gardon (1974)UNDEREMPLOYUENT RQUIEIBRIA 1195 augmented contracts will soften wage rigidity somewhat but affect little else, Of particular interest for this paper is whether these agreements will react 19 2 decline in product demand by reducing, primarily, hours worked rather than employment and thereby prevent (19) from holding. One reason we do not observe arbitrarily short work schedules*? has to do with the well-known tendency of people to prefer Lumpy leisure streams over entirely smooth ones, for example, long weekends, paid vacations, ere. Suppose, for instance, that labor preferences over alterna tive consumption and leisure streams are represented by monotone rransformations of the additive function we, ) = ED fate + ea, (24) where the expectation is taken with respect to the joine distribution. 46sq. Sy 04 Sp) of states of nature over the (presumed exogenous) duration of the contract. Then the tastes for lumpy leisure streams would be reflected in the requirement that &(f) be concave over some, interval (0, lg) and conven over (fo, 1). And work schedules which con- sume nearly all or nearly none of the leisure available to workers will bear prohibitively high “overtime” or “undertime” costs. The latter type of premium simply reflects the fact that it takes 4 very high hourly sage (© persuade people that they must abandon the comforts of home for 1 hour's work at the factory ds Contractust Unemployment bnvolunter? ‘The rest of the conclusions drawn front theorem 1 ate unexceptional, perhaps to the point of fomenting suspicion that they could be acrived at via the familiar auction model with some restrictions on labor mobility Indeed, the periect enforceability of contracts. an assuaptor uncom foreably close to indenture, docs guarantee that whoever is laid off when, (28) prevails will be unable to seek employment outside the industry. Furthermore, if interindustry Tabor mobility is low because of sy, moving costs oF industey-specific skis, some (voluntary) unemployment will occur in states for which a condition remumsceat af (19) holds, that is, PFU) om; (iii) MQ) > mf YEE LK, 20) fiv) ¥Q) 4 Las 2+ K, F(2) + Oasd + 00; &) ALY @LTMO)] = [TQ] YA e (K, «), 29 Casual enepricis ceinforcs this belt ao the extent that ii easier, ols costly, to seute information on se reputation and pest personel palsies affirms than on the ployentor record of st vers ighly paid employees SThe proot, whic eaentially valves characterizing te salotion of too siultansoue nonlinezr equations in two unkown, s available on request from the suchos,UNDEREMPLOVMENT EQUILIBIREA 1197 ya) mF Mea) we Mia, Fig. 2.—Emplayment schedules of optimal contracts Furthermore, the optissal contract 5° = {i*, x®(j)} satisfies (vi) Bagot) = at = ptt) < ws (vii) wt = TA): (vii) Hao) = [ult] — ajja wt) > Os (ix) at Gs) = FLe(w*) ito] for s < Fea = M(A} for s > FON; (x) 2*(5) is continuous, nondecreasing in 4, increasing in 4 for 5 < P(A), and decreasing in 4 for s > YA]; and (xi) wt(s) = mf at somes = < min {6 F(A} This theorem argucs that the wage associated with the underemploy- ment contract exceeds its fullemplayment counterpart, and so does work force size, The first of these two departures from full-ernployment norms is necessary t compensate laborers for a positive probability of unemployment; the second one allows employers to shift output from relatively unprofitable to relatively profitable stares of nature. As A rises relative t0 both the product price schedule, p(s), and the value of leisure, it becames more “expensive” for the firm to maintain a risky, that is, steeply sloped, employment schedule. Thus, the under employment and. fall-employment portions of n*(s) shift to the right and lefi, respectively (fg. 2a}, the critical state ¥(A} declines in valuc, and n¥(z} cores closer 10 2 fallkemployment schedule. Workers in active employment are paid a wage rate?? which exceeds the expected value of labor's marginal product by a marginal under- employment preiniom (uw). A comparison with equation (3, shows that 22-There coms te be no relationdhip between che fullemployment wage rate and the ‘expected marginel product of labor under the optmal contract,1298 JOURNAL OF POLITICAL ECONOMY his a smallee function than @, that is, (0) — Yu) Olorallw ok 26) The difference arises from and if being the marginal underemployment premium in the class D of feasible coniracts with xed labor pool mj and in the class F of all feasible contracts, respectively. If employment below mf is called for in some states, class D must inevitably present greater unemployment risks than class F: the option of varying the size of the labor force is available only in the latter. For fixed tastes, technology, and population, the equilibrium value of contracts is determined, in the usual manner (fig. 26), by the interaction of a Gxed supply M' = A of workers with economy-wide demand, MO = 3, Mii) for “work force,” where t indexes industries, Contracts wersus Auction?» We return now to the issue of whether the employment (and output] pattern in a labor contracts economy is identical to, or varies systemate ically from, what would prevail if labor services were sold in instantaneous auction, With unlimited labor mobilicy, the later type of market organiza tion is Pareto efficient and, hence, must generate at least as high a volume of emplayiment and output as 2 market for contractual bargiins A less vacuous comparison would recognize that low mobility is ane of the primary reasons long-term attachments arise between workers and firms, Reduced ta its barest essentials, the issue is: given a Labor stock of se workers (in per fitm terms) without job opportunities outside ehe ine dustry, how dees the outcome of an auction market compare with che terms of the optimal cantract? Let 0,63), n°), nt(5), and n,(6) be the wage, labor supply, labar demand, and employment schedules, respectively, in the auction market. Define 3° from PF) =e (262) and note that & = z(é). Next let the function gf, m] solve Algh"(m) = tm). @7) Now one can show?* Lemma 6: The function tw, m) is unique, continuously differentiable, decreasing in 2, ineredsing in m, and bounded above by | and below by @, and is such that (i) 5° = ¢(&, 4) solves (26a): and (Ii) glé 4) > ¥(2}, where ¥(A) is defined in chesrera 2 £1 Lam indebted to Milton Friedman for provoking this section, Pe NUZL azplies here, 00.UNDEREMPLOVHENT EQUILIBRIA 1199 s pts.) ota) nde) 54 qth wl es, (ke) TOON dn = Fr. 3 Auction sare equa Now since Ae) = Pts) 6))5 6) = ge itaylsy > a = Vifw,{s) < 4, it follows from lemma 6 that is) = PUP whe) = IES << lls 1) a) =H > = pf lw) ifs > ath, a) ‘These relations are graphed in figure 3 In the contractual market, assume that a fullaemployment contract paying the lowest possibte wage is profitable an average, that is, FB) Ep > k 0) ‘The optimal agreement will be of the form 5f = (f"u}hips #) it Ble) > ALF UIE) (1) holds; otherwise, theorem 2 is in farce, with the additional proviso that M2) =p (32) Let wt, nt(s), and 2, be the wage rate, employment schedule, and expected wage income under the optimal contract and (et a be the ex- pected wage income in an auction market. Note that alk)gth, w) + Sf (w)pisids, (33) sha Now, if (31) ix true, then Hence, wef = FUE and x*(s) = p for all s f LF tulple) = stelle < 0 (a) lo1200 JOURNAL OF POLITICAL ECONOMY ns) Fia, 4—Camparative emplayment and, from (29), (6) leh > Hs < shh 19 (98) If, on the other hand, (31) does not hold, then we know that zo") < and 0% € (4, ag); hence, part (ix) of theorem 2 suggests that (35) is equally truc in this case also, Note, further, that in this case again wh. = WER Ye = w*(2 — K)jfule*) — XK], which is readily shown to be a decreasing function of te* for all to! in che relevance range (f wy) and all> K Hence, ye < wTU. — Kojfelwep — KP = wf = SEP =m, (B4b) Observe also frore theorem 2(¥) and lemmra 6 that FO) = af TA), MUA = atu, we) < 2 (he) ‘This completes the proof of Theorem 3: (i) i, < dys and Gi) m6) Stel Ys, < Fors < glk ws where (ii) ef wl > FQ). As one might expect, labor markers generate riskier employment schedules (fig. 4) and larger expectations of wage income when they are organized as auctions vathor than as contract exchanges. ‘VI, Summary and Conclusions ‘This paper examines a simple industry with demand uncertainty which motivates risk-neutral firms to act both as employers and as insurers of homogeneous, risk-averse laborers. The resulting contractual agreeCNDEREMPLOYMENT EQUILIBRIA tar ments turn out, in their simplest form, to be more likely to specify full employment tee more ofthe following conditions prevail: small variability in product price, above-average economy-wide labor demand, bighly riskeaverse workers, smali unemployment compensation, and highly competitive product market. Otherwise, it may be optimal to lay off, by random choice, part of the work force during low states of demand In such “underemployment” contracts it is profitable, under certain assumptions, to eliminate wage fluctuations, but not employment status, as a source contributing to wage income variability. If the aber force is indusery specific, conicactual organization of the labor market will result in the volume of employment being no lower in any state, and higher in some states, than that associated with a labor auction, ‘The employment pach generated in 2 contractual labor market is aot mnuich different from the ane which prevails ina spot auction market; because of rigid wages, however, reductions in employment appear 2s layoffs rather than as voluntary withdrawals of labor services Some extensions of the work reported in Sections IT-V are worth contemplating, The firsc involves dropping the assumption of perfect homogeneity of the labor force and finding out what sort of wage di butions, layof provisions, and participation rates are consistent with labor market equilibrium when difforenc contracts are offered to persons of varying technical skills, tastes, asset positions, and home production ‘opportunities. A related paper of mine (Azariadis 1975) examines the incidence of unemployment in a technology with two asymmetrically substitutable skill grades of labor. That paper finds that the trade-offs between wage income and the expected rate of unemployment embodied, in optimal contracts clearly favor skilled workers over common labarers. Anozher extension, passingly discussed in Section IV, is xelaxing the restriction of contract enforceability, at least on employees. In a model with at least two industries whose product price ratio varies according t0 a wellsdefined stochastic. process, unilaterally enforceable. contracts may lead to a natural definition of (ternporary] vacancies and, in ail likeit- hood, to a softening of wage rigidity. Perhaps the most interesting area of application for the class of models treated in this paper arises when the assurnption of exageneity in the price level is dropped, Perceived disturbances of the price level relative either to its average long-term value or to the price of commodities in in- formationally segmented markets are at the heart of modern treatments of the Phillips curve.** Whether we can model price movements of this sort to lear something about wage rigidity in the large and, more generally, optimal escalator clauses in money-wage contracts seems to me an issue worth pursing. 8 Sce Friedman (1968), Phelps (1968), and Morsensen (1970) on the fale af mover seats in contemporaneous priee ratios and Lucas ane Rapping (1469) ane Leas (1972) fn the role of mavemen in interterporal one.1908 JOURNAL OF POLITICAL ECONOMY References Acro, BJ. Boys ia the Theory of Risk-Besring, Chicago: Markham, 197%. New York Times (November 26, 1972). Aaariadis, C. “On the Incidence of Unemployment." Res. Bian, Stulie (in press, 1975) Baily, M,N. “"Wayes and Employment under Uncertain Demand.” Rea, Ean, ‘Studies 41 (January 1974) 37-50 Barro, R. J., and Grosican, H. T.'“A General Disequilibrium Madel of Income ‘ni AEA Bl (March 1971): 82-93 . G. "Real Wages and Cyclical Variations in Employment.” Conadisn “Bean. 2 (August 1968): 353-74. Brainard, W., and Dolbear, F. T. “Social Risk and Financial Markets." A.E.R. 61 (May 1971): 360-73. Clower, R_ “The Keynesian Counter-Revalution: A Theerstical Appraisal.” Ta The Theory of Inoret Raver, edited by P. HL. Hahn and F. P. R. Brecbling, London: Macmillaa, 1966 Danforth, J. P. “Expected Utility, Mandatory Retirement and Jab Search.” Working paper, Univ. Minnescta, 197%. Friedman, M. “The Role of Manetary Polley." 4.2.2, $8 (Mazch 1968): 1-17, Gordon, D. F. “A Neo-lassical Theaty of Keynesian Unemployment.” £22 Jnguizy 19 (Docernbyer 1974): 431-59, Hicks, J. R. The Theory of Wages. Londion: Macenllan, 1932. Keynes, J. M. The Generel Theory of Employment, Interest and Mongy. Landon Macraillan, 1936, Knight, FH Ris, Uncertain and Prope, Boston: Houghton-Miffin, 1921 Laidier, 0, EB. W, ‘The Domced for Money Theories and Eetdence, Scranton, Peas, International Textbook, 1960 Leljonbutvud, A: On Kejnesizn Bzonamice and the Economics af Keynes. London: Oulord Univ. Press, 1958. Lucas, R. E. “Expectations and the Neutrality of Money." J. Kian. Thonsy 4 (April 1972) 103-24 Lucas, R.E., and Rapping, L. “Real Wages, Eroplayment, and Inflation.” PAE 72, pe. 5 (Qetaber 1969): 721-4, Melteer, A. HH, “The Domnand for Money: The Evidence from the Tite Series" JPET, 10, 3 (Sune (963): 219-46. Morcensen, 1D. “A Theory of Wage and Employment Dynamics." In Afra. ecaramic Paurdavons of Emplyment and Inflation Theary, edited by E. 8. Phelps et al. New York: Norton, 1970. Parker, J, Ba, and Burton, J. F Je. “Vatuntary Leber Mobilicy ia the U.S, ‘Manulacturing Sector” Tn Retdinas in Libor Marke! Anaissr, edited by J. B. Burton ev al, New Yark: Holt, Rinehart & Winston, (971 Patinkie, D. Mong, Inarest, and Prices, New York Harper & Rew, (O56 (2d ed., 1965). Phelps, E. 8. “Monay Wage Dynaruies and Labor Market Equilibrium.” J.P.E: 76, 20. 4, pt. 2 (August 1968) = 688-711 Prati, J. W. “Risk Aversion in the Small and in the Large.” Exowmetrica 92 (January 196): 122-36 Stighita, J. “Incentives and Risk Sharing in Shareeropping.” Ren, Bann, Studer 41 (April 1974): 219-56, ‘Tobin, J. “Money Wage Rates and Employment.” Tn The New Beonomies, edited by SIE, Harris. New York: Knapf, 1987. ——. “Inflation and Unemployment." 4.E.R. 62 (March 1979): I-14)