A General Equilibrium Entrepreneurial Theory of Firm Formation Based on Risk Aversion Richard E. Kihlstrom; Jean-Jacques Laffont The Journal of Political Economy, Vol. 87, No. 4 (Aug., 1979), 719-748. Stable URL hitp:/Mlinks jstor-org/sici?sici=0022-3808% 28 197908% 2987%3A4%3C719%3AAGEETO%3E2,0.CO%3B2-B ‘The Journal of Political Economy is currently published by The University of Chicago Press, ‘Your use of the ISTOR archive indicates your acceptance of JSTOR’s Terms and Conditions of Use, available at hup:/www,jstororglabout/terms.hml. ISTOR’s Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use ofthis work. Publisher contact information may be obtained at hupulwww.jstor.org/journals‘uepress html. Each copy of any part of @ JSTOR transmission must contain the same copyright notice that appears on the sereen or printed page of such transmission. STOR is an independent not-for-profit organization dedicated to creating and preserving a digital archive of scholarly journals. For more information regarding JSTOR, please contact jstor-info@umich edu hupslwww jstor.org/ ‘Tuo Jan 6 10:03:40 2004A General Equilibrium Entrepreneurial Theory of Firm Formation Based on Risk Aversion Richard E. Kihlstrom University of Mins, Urbana-Champaign Jean-Jacques Laffont Laboratoire dEconometri, Ecole Plyechnique, Paris We construct a theory of competitive equilibrium under uncer: using an entrepreneurial model with historical roots in the w Knight in the 1920s. Individwals possess labor which they c 1s workers toa competitive labor market or use ts en srs have saece: running a firm, All entrepre sand receive all risk averse ind ers while the less rs nirepreneurs. Less risk averse entrepreneurs run larger firms and economy-wide increases in risk aversion reduce the librium wage. A dynamic process of firm entry and exit is stable, The equilibrium is efficient only ifall entrepreneurs are risk neutral Inefficiencies in the number of firms and in the allocation of labor to firms are traced to inefficiencies in the risk allocation caused by institutional constraints on risk trading. Ina second best sense which aecounts for these constraints, the equilibrium is efficient to the same ms. In the equ I. Introduction The recent work on the economies of uncertainty has failed to achieve general agreement to the goals which motivate firm behavior under escarch support from the National Science Foundation under grant, SOC-76 11588 is gratefully acknowledged. Much of this work was completed while Kiblstrom was a visitor a the Laboratoire d'Eeonomettie, Lcote Polytechnique. Their support for ihn ly appreciated. Finally, we also would like to thank Jean Michel art Glenn Lary, Robert Leas Ed Preset, Servi Rosen ae sschein for helpful comments, al f Pal Foe HP, ot 4 TRE The Une of chap, es SeoNTN TU. 24 79720 JOURNAL OF POLITICAL ECONOMY unce ng firm decision making which have the existing literature include expected jon and expected utility of profit maximization as well as maximization of the rket value. Difficulties with each of these criteria have led to a discussion (‘Symposium on the Optimality of Competitive Capital Markets” [1974]) of the conditions. under which there exists a criteria for firm decision making which achieves unanimous approval of stockhold have also led 10 the study of other more sophisticated criteria for firm decision mak- ing. The paper of Dréze (1974) is one in which this latter approach is taken, Fach of these subsequent approaches ha mity can be achieved only in limited technological circumstances. Similarly, the equilibria of Dreze are not always efficient in the “second best” sense of Diamond (1967) In this paper we construct a competitive general equilibrium theory of the firm under uncertainty which is based on an en al ‘model having its historical roots in the work of Knight (1921). The entrepreneurial model permits us simultaneously to use the expected ¥y maximization criterion and to provide a justification for its use. This is accomplished by assuming that for each firm there is an expected utility maximizing entrepreneur who makes decisions for the firm, Furthermore, the model uses a free-entry assumption to endogenously determine the n 's and the identity of the run them, It also permits u ber of fir entrepreneurs whi vidual characteris preneurs, operating a risky firm or working for a riskless wage. There are, of course, many factors which should influence this choice. The most important ones would epreneurial ability, labor skills, autitucles toward risk, and initial access to the capital required 10 ¢ firm. The present paper focuses on risk aversion as the determinant which explains who becomes an entrepreneur and who works as a laborer. The equilibrium which is shown to exist has the property that less risk averse individuals become entrepreneurs, while the more risk borers. In addition to providing an explanation for the identity of entre- preneurs, the entrepreneurial model can also be used to study several issues of traditional interest 10 economists. One of these is the process of firm entry and exit Specifically, using the model described below, it is possible 10 zee the dynamics of firm entry and exit in a general equilibrium context, This can be done using a formalization of a vatonnement process which is analogous to that commonly used to study the stabil- averse work asTHEORY OF FIRM FORMATION 7a ity of competitive equilibrium. While our stability analysis is less complete and more special than the analysis in the stability literature for competitive equilibrium, it nevertheless introduces an elemen specifically firm entry and exit, which this literature was unable to, fe. Furthermore, this element is introduced while retaining I equilibrium framework and the basic price-adjustinent process. In our general equilibrium process, as in the titonnement -¢ equilibrium literature, prices (in our to (labor) market disequilibrium by rising when is excess supply. Earlier formalizations of the entry-exit process, specifically, those in Quandt and Howrey (1968) and Brock (1972), were par librium models. They were also based on formaliza- tions of the adjustment process which, while similar in spivt, differed in detail from the titonnement price-adjustment process used in the competitive equilibrium framework. For example, in the papers by Quandt and Howrey and by Brock, the dynamic variable is the number of firms in an industry. The industry grows when profits (or excess profits) are positive; it contracts when profits are negative ‘Another traditional question which can be investigated using the entrepreneurial model concerns the determinants of the distribution of firm size. Specifically, an entrepreneur's attitude toward risk can be related to the size of the firm which he operates. While it might be conjectured that more risk averse entrepreneurs run smaller firms, this is not always true. However, it does follow when the production function satisfies certain conditions which are spelled out in theorem 3 below It is also possible to use this model to study one determinant of the distribution of income between workers and entrepreneurs. Specifically, it can be shown that, under certain conditions, the equilibrium wage level would be depressed if the economy's popula tion be ore risk av Finally, itis possible to i te the efficiency of the eq of the entrepreneurial model, In general the equilibrium is inefficient and the inefficieney takes three forms: risks are n are operated at the wrong levels, and there is an inappropriate number of firms. It is shown, however, that all of these forms of inefficiency occur because there are institutional constraints em bodied in the model which prohibit an efficient allocation of risks when entrepreneurs are risk averse. This is seen in two ways. First, the equilibrium is efficient if, in equilibrium, all entrepreneurs are risk neutral. Second, we follow Diamond (1967) and Radner (1968) and investigate the efficiency of equilibrium in a second best or “limited sense which permits a less than completely efficient allocation of risks Although the “limited efficiency” approach taken in this paper is injae JOURNAL OF POLITICAL ECONOMY the same spirit as those adopted by Diamond and by Radner, it employs a different concept of limited efficiency than the ones they employed. Thus we accept the specific institutional constraints im- posed by our equilibrium model on the distribution of risk and ask only that, given these constraints, all other decisions be made efficiently (Pareto optimally), The constraints on risk trading im- posed in taking this approach are, in fact, stronger than those intro- duced by Diamond and by Radner. It is possible, however, to show that if, in defining limited efficiency, these constraints are imposed, then the equilibrium is efficient in the limited sense. Because all of the inefficiencies which may arise in an equilibrium can be aced to the institutional constraints on risk trading, reasonable to conjecture that the efficiency properties of the equilib rium will be substantially improved by the introduction of at least some market opportunities for risk sharing among entrepreneurs and between entrepreneurs and workers. The introduction of a stock market in which the entrepreneur can raise capital for the purpose of financing his input purchases would be one way of providing addi- tional opportunities for risk sharing. Sharecropping arrangements provide another device by which risks are, in Fact, often shared. This is especially true in agricultural economies. The present paper does not investigate the issues which arise when either of these risk sharing. possibilities becomes available. In a subsequent paper (Kihlstrom and Laffont 1978), we have, however, succeeded in studying these exten- sions of the entreprencurial model discussed here. The emphasis there is on the stock market as a device for risk sharing. It is lly shown that the introduction of a stock market does, i ‘sult in equilibrium allocations which are efficient in a stronger hat considered here. Specifically they are efficient in the yond. This paper conclud discussion of their relationship to Knight's above-ment preneurial theory with a brief summary of our results and a mned entre II. The Model The set of agents is identified with the interval 0,1]. If a € [0,1] individual @ has the von Neumann Morgenstern utility function (La) where I represents income, and J € (0,%). For all J = 0, the first and second derivatives uw; and uy, exist and are continuous. The ‘marginal utility u, is positive and noninereasing, that is, 4, = 0. Thus all agents are risk averse or indifferent to risk We also assume that the Arrow (1971)-Pratt (1964) absolute riskTHEORY OF FIRM FORMATION 723 aversion measure is nondecreasing in a. More precisely, if « exceeds B, then agent a is at least as risk averse as agent B in the sense that a for all I € (0, 2), Each agent can become an entrepreneur and use without cost a technology defined by a continuous production function y = g(L.x) wherey 2 0 is output, L 2 0 isthe labor input, and x is the value taken by a nondegenerate random parameter with support [0, ¥], 0 < Re tn, The marginal product g, is assumed to be continuous and positive on (0,+%) x (0, x]. The second derivative is continuous and nonpo: tive on [0.+2) x [0, ]. Thus g exhibits nonincreasing returns to scale for each x. In addition, g(0,x) = g(L,0) = 0 for all x € [0.x] and L € (0,4), while g(L.x) >'0 on (0.4%) * (0,3) ‘A variety of interpretations of the random variables & is possible. In all of these interpretations, the stochastic distribution of & is as- sumed to be the same for all firms. On the one extreme we can assume that the random variables which determine the output of each firm are stochastically independent. Atthe other extreme they €an be perfectly correlated. In this case, not only isthe distribution of & the same for all firms, but the same random variable & influences the output of all firms. Intermediate cases occur when the %'s are correlated but not perfectly correlated. In each of these alternative interpretations, all \ividuals are assumed to have the same beliefs about the distribu- of &, that is, the distribution of & is objective. ‘The price of output is | and labor is hired at a competitive wage w. It is assumed that the demands of entrepreneurship preclude addi tional work by agents who choose to operate a choice. They can become entrepreneurs ai income or they can work and receive the market wage w. If an individual becomes an entrepreneur and employs L workers he will receive profits equal to g(L.3) — wh. @ To avoid the difficulties associated with the problem of bankruptc we assume that all individuals begin with A units of income and that they are unable to hire workers who cannot be paid if ® = 0. Thus L. must be less than or equal to A/w An individual who becomes an entrepreneur will choose 10 employ74 JOURNAL OF POLITICAL ECONOMY L(wa)_ workers where L(wa) is the L value in [0, Ako] which Bu(A + g(L.3) - wh, a) ) Our assumptions on u and g guarantee that L0,a) exists. If either uy <0orgy, <0, then L(e,a) will be unique. When entreprencur a faces the wage w and employs Lisa) workers, his profi are random and equal to, a) ) If the wage is w, agent a will choose to be an entrepreneur when A(w.a) = g(L(w.a), 8) = wh. Eu(A + #w.a), «) =u(A + wa). () He will be a worker at wage w if Eu(A + Hwa), a) 5 (A + wv, 0), ©) and he will be indifferent if the equality holds in (5) and (6) Equilibrium is reached when the labor market clears. At the equilibrium wage, the labor demanded by all agents who choose to become entrepreneurs equals that supplied by agents who choose to enter the labor market Formally, an equilibrium isa partition {4,T} of [0,1] and a wage w, that is, a pair (A,0}.0); for which (E.1) labor supply equals labor demand in the sense that Js Lowegutda) = wi) where wis Lebesgue measure and (£2) for all @ € A (5) holds and for ae (6) holds, IIL, The Existence and Uniqueness of Equilibrium We can now prove that an equilibrium exists, The first step is to define w(a), the certainty equivalent wage which makes agent indifferent between the two activities—work and entrepreneurship. Formally, w(a) is defined by cu(A + i(w(a), a), a) = u(A + w(a), @) a The properties of w(a) are established in the lemma which follows. These properties will permit us to describe the structure of the equilibrium in a way which simplifies the existence proof. Further interpretive remarks follow the formal statement of the lemma,THEORY OF FIRM FORMATION Lemma Assume that for each /, r(Z,a) is an increasing function of a! Also assume that either gj, < 0 or uy, < 0. Then: i) For each a € (0,1) Eu(A + Hwa), a) — (A + wa) isa continuous monotonically decreasing function of w ii) w(q) is a well-defined function of a, that is, for each a € {0,1 w(a) exists and is unique. In addition w(a) > 0. i) If w > (<) w(a), then Eu(A + #u.a), a) < (>) pA + ). iv) Ia > B, then w(a) < w( 6). ¥) WB > (=) a, then Eu(A + (w(a), B), B) < (>) u(A + wa), 8) vi) IF 0 w < mB), then Liu) > 0. This is true, in particular, if w = w(a) where a > B. Remark 1 Result iv asserts that more risk averse individuals are induced to become workers at lower wages than less risk averse agents. In order to interpret result v, note that agent @ will be the marginal en preneur if the equilibrium wage is w(a). Result ¥ asserts that all indi- Viduals who are more (less) risk averse than the marginal entre- preneutr will be workers (entrepreneurs). This result implies that in any equilibrium, there will be a marginal entrepreneur d for whom w(a) is, the equilibrium wage. The set of entrepreneurs A will be the interval {0.a and the set of workers will be (11. The problem of finding an equilibrium then reduces to the problem of finding a marginal en- trepreneur é for whom E.1 holds when w = w(@), A= (al, and P= (G11, that is, for whom f.'L (w(a)a)u(da) = 1 = & PROOF —{i) The assumptions mace about w and g guarantee that Eu(A + g(L.x) — wl, a) isa strictly concave continuous function of L, and a continuous function of w To prove monotonicity, note first that for each nonne} monotonicity of w implies that re L., the Eu(A + (La) ~ wha) ’, Maximizing over I. on each side of inequality (8) implies the inequality icly increasing the sic inespities init, ana ware replace by wei inequalities726 JOURNAL OF POLITICAL EeoNOMY: Eu(A + (wa), a) +a) Emax Eu(A + g(L3)—w'La) — @) Eu(A + atw'a),a) when w >", Thus Eu(A + #(w,a), a) is noninereasing and Eu(A + F(w.a), a) — u(A + wa) is monotonically decreasing ii) Iv is easily seen that Eu(A + #(w.a), a) ~ u(A + w,a) > 0 when 0. If, on the other hand, w is large, then Osxst and equation (6) will hold. Because of the continuity established in i, the intermediate value theorem implies the existence of a positive wage w (a) which satisfies (7). The monotonicity established in iin the uniqueness of w(a). Monotonicity also implies inequality ii. Figure 1 illustrates the situation. Eu [A+F(wa),a] la u (w, a)wia) w(8) w or-te7 Fre. iv) We use the fact that w(a) is the certaimty equivalent of #(w(a), 4a). We also define w(a,8) to be the certainty equivalent of #(w(a), a) for agent 8. Pratt's (1964) theorem 1 is now applied to prove that 8 > () e implies w(a,8) < (>) w(a). The monotonicity of w(w,B) inw then, guarantees that Eu(A + a(w(a). a), B) = u(A + wla,p), 8) <(>)u(A + wa), B) ay when B > (<) a Now note that, by definition of #(w,8), Eu(A +7(w(a), 8),8) 2 En(A + #(w(a),a).B)- (12) When B < a inequalities (11) and (12) combine to yield Eu (A + #(w(a), 6), B) > u(A + wla),B) a3) Figure 2 illustrates what is easily proven; that inequality iv is a conse- quence of inequality (13), the equality defining w(B), and the fact that Eu(aw,B), B) ~ u(w,B) decreases monotonically in w. ¥) Inequality v follows immediately from iii and iv vi) Since iii implies that B) B) >u(A +w,B), if wip) > w, #(w, B) = g(L(w, B), 8) — wh(w, B) must exceed w > 0 with positive probability. This is impossible if 1.(, 8) = 0. ||728 JOURNAL, OF POLITICAL ECONOMY In the discussion of existence, the analysis is restricted 10 cases which satisfy assumption A: u(/,a) is everywhere a contin ion of a THEOREM 14—Assume that for each I, r(Z,a) is a nondecres function of a, Also assume that either gi, < 0 or uy < 0. Under assumption A an equilibrium exi PROOF.—Undeer assumption A it can be shown that our assump: tions guarantee that L(w.a) and w(a) are continuous functions of aon 10,1]. Thus for each w € [w(0)s0(1)] and & € [0,1], f,"L(o,a)da exists. We can now find an a such that sf JO (uta), ada = 1 = 6 (ay Note that JL (w(a), adda — (I~ &) is a continuous function of & which is negative when & = 0 and positive (by viof the lemma) when = 1. The intermediate value theorem implies the existence of an satisfying 14. Now we ean define {AD}, w) = ({L0.4°), (4,1, we) (15) {AP}, w) = ({L0,6*), [4% 1], we). For these entrepreneur, worker, wage combinations v of the lemma implies that condition E.2 holds while E.1 reduces to (14). | | The next theorem gives conditions under which the equilibrium is, unique: THEOREM 2.—Assume that for each 1, r(J.a) is a nondecreasing function of a. Also assume that either gi, < 0 or uy <0. If, in addition, L(u,a) isa decreasing function of w, then the equilibrium is unique. PROOF —Because of the lemma, and for reasons discussed in mark 1, the equilibrium occurs at an & for which (14) holds. In addition, the lemma implies that Lo.) > 0 for all a and all w * x(a), and that w(d) = w(d’) if @ < @, Then since L(w,@) is a decreasing function of w, @ < 4’ implies [eCora ye) - [eo r.0)ta + [Hla de> [101A ete "This theorem ean he proved without assumption A. The assumption is made solely to permit a simple existence proatTHEORY OF FIRM FORMATION 79 Thus labor demand at w(@), [,'L(w(a),a}da, is a strictly increasing function of &. Furthermore, labor supply at w(4), (1 ~ @), is a str decreasing function of @ Therefore excess demand at w(é Jy'L(o(a),a)der ~ (1 ~ 6), is a strictly increasing function of and there can be only one é at which excess demand can equal zero, that is, at which (14) can hold Conditions under which L(u,a) is a decreasing function of the wage w are discussed in remark 4 at the end of the following section on ‘comparative stat IV. Comparative Statics Having established the existence of an equilibriu to study its properties. Specifically, we can first ask how a firm’s size, as measured by its labor demand, is related to the risk averseness of the entrepreneur running the firm, It might be expected that more risk averse entrepreneurs operate smaller firms, that is, use less labor than less risk averse entrepreneurs. Theorem 3 gives conditions under which this expected result obtains. The conditions require that \ge inx must affect output and the marginal product of labor in the same way; if an increase in x raises output it must also raise the marginal product of labor, One important special case in which this condition holds occurs when the uncertainty enters multiplicatively i is now possible g(iL.x) = xh(L)S (16) THEOREM 3.—Assume that L(w,a) < Abo. If g(L.x) and gu(L.x) are both monotonically increasing or both monotonically decreasing functions of x, then L(w,a) is a monotonically decreasing function of The proof is essent is not reproduced. We can now ask to what extent itis possible to describe the influence of individual atitudes toward risk and of technological parameters on the equilibrium. In general, not much can be said about the effect of these eters on the number of firms, a variable of particular interest. But for the purpose of studying the distribution of wealth Detween workers and entrepreneurs it is important to know how these parameters influence the wage. What ean be shown is that, under certain reasonable conditions, an increase in individual risk aversion reduces the wage. lly the same as that given in Baron (1970) and 2 This isthe case considered by Baron (1970), nv Baron's paper i interpreted 38 price. Equation (16) is alo included in the class of cases studied by Diamond. In Diamond's terminology, (18) represents case of stochastic constant returns to scaleJOURNAL OF POLITICAL ECONOMY 4.—1F (iin equilibrium, all entrepreneurs are identical, (i) either gi, <0 oF ty < 0, (il) g(L,x) and g,(L.x) are both monotoni- (or decreasing) functions of x, and (iv) L(w,4) is an 1d a decreasing funetion of w, then an increase in the Arrow-Pratt absolute risk aversion measure r(J,@) for all , lowers the equilibrium wage. Remark 2 The intuitive basis for this result is as follows. Since, in equilibrium, workers are the most risk averse individuals, an economy-wide in- aversion increases the supply of workers and this tends to lower the wage. This tendency is reinforced by demand changes implied by theorem 3 which applies because of assumption ii Specifically, theorem 3 implies that an increase in the entrepreneurs’ aversion to risk reduces the demand for labor; PROOF.--If, in equilibrium, L(w@) is an interior solution, the first- order maximization condition for the marginal entrepreneur is Bu(A + g(L,) = wh, dg, (L.%) = Eu(A + gk) — wh, aw (17) where L = L0w.4) We also have (A +0.) = Eu(A + gil ~ wh, @) as) at = Lewd) The equilibrium conditions (17) and (18) imply relationships be- tween L and w which are described in figure 3. The relationship implied by (17) is downward sloping because of assumption iii. As indicated, (18) implies that w is a function of L which reaches its minimum when it intersects the line defined by (17). This is proved by differentiating (18) implicitly to obtain wi(A + glad) — wh, a) [gull dL UA + wad) + Eu + gL.) — wh, QL dt =u) ag) “The differentiation is justified by the implicit function theorem be- cause the denominator in (19) is positive. The numerator is zero wher (17) holds. The second order sufficient condition for the entre- preneurial maximization problem is satisfied because of condition ii ‘Thus, as reflected in figure 3, the numerator in (19) is positive (nega- tive) and dwidL is negative (positive) when L< (>) Ltw, @). ‘ow suppose that r(,4) increases for every /, then theorem 3 above guarantees that L(w,@) is lower for each w. Also, reasoning similar to that employed in the proof of the lemma guarantees that for each L,THEORY OF FIRM FORMATION 731 a-{iL,w) which sotisty (17)} {(L.w) which sotisty (18)} | op-168 Fie. 3 the wage level w at which (18) holds is alo reduced. ‘Thus the + se affects the relationships between Land w implied by (17) and iS shown in figure 4. As a result the equilibrium wage must decline. | | We can now make several observations whic marks. we formalize as re Remark 3 Asin is always an increasing function of w. lar proofappliesif L(w Remark 4 There are several important cases in which L(w,@) is indeed a de- creasing function of w. These occur when either (a) r(/,4) isa constant nction of f, (b) g(L.x) and gi(Lx) are both increasing (or both decreasing) functions of x and r(/,@) is a nonincreasing function of 1, ‘or (©) gi(L.x) satisfies (16) and — tut + 1a) ut +1, 8)738 JOURNAL OF POLITICAL ECONOMY. Equation (17) oP-iss forall. The condition imposed on u(-,4) ine asserts that when w(A + 1, @) is considered as a function of it has Arrow-Pratt relative risk aversion less than or equall to one. PROOF OF REMARK 4—Implicitly differentiating (17) we obtain wy (A + glk) ~ wh, agile) ~ w) _ Eula + giL.i) — wha) 20 where D = Euy(A + gibt) ~ wh, aigu(L.k) ~ wl + Bu(A + g(L.t) — wh, @gu(L,x). The second-order condition for the entre- prencur’s maximization problem (which is implied by condition ii of theorem 4) guarantees that the implicit funetion theorem applies to justify the implicit differentiation. This condition also asserts that the denominator in (20) is negative. In general, the sign of the numerator in (20) is ambiguous since the first term is ambiguous. (The second term is negative.) In case a, however, the first term is 4) [glk (A + gtL,8) ~ wha) [gi(L.8) — w] LBuy(A + g(L8) ~ en which equals zero because of th case a, the numerator is negati firsvorder condition (17). Thus in as is dL.idw.THEORY OF FIRM FORMATION, 733 In case b, the first term can be shown to be nonpositive by an argument similar to that used by Baron (1970). Tn ease ¢, the fact that A”(L) = 0 can be used 10 obtain W'L)Le — wh) SHLIE ~ wh (2) and ~uy(A + Wy wh @)W’U)LS = wh] — uy (A + WLR ~ wh @)Ih (LY ~ wh) When this inequality is combined with ¢, the numerator in (20) is negative. | | (23) Dynamics In this brief section, we consider the stability of a Gitonnement ad- justment process similar to that used in studying the stability of ‘competitive equilibrium. In this process, the wage is assumed to adjust to labor market disequilibrium by rising when there is excess demand and falling when there is excess supply. Specifically, elf Lat, alde ~ (1 wit) en where isa differentiable increasing function such that (0) where d(w) satisfies the equation Eu(a(w, lw), dw) = ulw, aCe). (25) Oand We define Vee) = (af Le. ante = 11 ~ awn 26) to be the Lyapunov function, Thent A v(wio) = 26 (P.0 anda — (1 - aw) x {[Leo@a(wn)) +1 en “f Now 6! > 0 by assumption, and a’ (sw(0)) is negative because of the leimma. Thus if ./ < 0, didt V (o() isnegative and w(t) converges to the equilibrium wage. lawn) a Car} w(t ai SAF we assume that yal ag exist ane! are continuous, repeated application of the iamplccfametion theorevsinplis that de) is 4 iferenniable functian of734 JOURNAL OF POLITICAL ECONOMY These results are summarized in the following theorem. THEOREM 5.—Assume that for each I, r(I,@) is a decreasing func- tion of a and that either gi, <0 or uy, <0, Also assume that and ty exist and are continuous. If L,.(w,a) exists and is negative, then w(0) converges to the unique equilibrium wage. Remark 5. the standard explanation for firm entry and exit, which does not admit the possibility of uncertain profits, reductions (increases) in profits caused by falling (rising) demand or increases (decreases) in cost result in exit (entry). In our entrepreneurial model demand. changes are not explicitly considered and cost changes are introduced, by wage changes. In addition, changes in the return to nonentre- preneurial activities, specifically labor, also cause entry or exit. Again. these changes are embodied in wage changes. The fact that returns to nonentrepreneurial activities influence firm entry and exit is a reflect jure of our formalization. In this framework, jon to enter as entrepreneur or exit to become a worker is made after the ex- pected utility of the random profits available to entrepreneurs been compared to the utility of the nonrisky wages earned by workers. In the formalizations of Quandt and Howrey (1968) and of Brock (1972), firms decide to enter if (excess) profits can be made, This is appropriate when there is no uncertainty and no opportunity cost 10 entry other than capital costs. In our model both of these complica- tions are present. Profits are random and the opportunity cost of becoming an entrepreneur is lost wages. Remark 6 Since the model of the adjustment process studied here is analogous. to that employed in the literature on competitive equilibrium, it is subject to the same criticisms. Specifically, the dynamic wage change ‘eqiiations are not explained by an underlying model of ma ig behavior. In the paper by Smith (1974), the dynamic equations which describe the process of firm entry and exit are explicitly obtained. from a maximization model VI. Efficiency of Equilibrium In this section, two concepts of efficiency are studied. The first is unconstrained Pareto optimality in the sense of Arrow (1964) and‘THEORY OF FIRM FORMATION 735 Debreu (1959). The second is a constrained version of Pareto opti- ality in which the institutional constraints on risk trading implicit in our concept of equilibrium are imposed on all allocations. ‘The rea- sons for studying constrained optima will be suggested by the analysis, of unconstrained optima. We will show that because of the institu- tional restrictions embodied in our equilibrium concept, asking for rained optimality is, in general, clearly asking for too much, There are, nevertheless, interesting cases in which an equilibrium is in an unconstrained sense. In addition, itis possible to specify ture of the unconstrained inefficiencies. Before proceeding to the formal discussio introduce special assumptions which are employed to nalysis of unconstrained efficiency. Specifically, we now is the same random variable for all firms, that is, that the random variations in the firms outputs are perfectly correlated. ‘This assu tion will be sufficient to permit an intuitive explanation of the inefficiencies occurring in our model. Furthermore, a general treat- ment would take us beyond the scope of the paper. The reader should note however that this assumption is used only in the discus- sion of unrestricted efficiency. In the subsequent discussion of re- stricted efficiency, no assumptions are made about the dependence or independence of the returns to different firms. Asa preliminary to the formal discussion, we define an unrestricted feasible allocation as a specification of Pand 4 and of funetions v:A—> 1y(-,x):(0,1] + (0,2), for each x, which satisfy the conditions { rladutde) = wi) (8) [/ stxrmiday = | e(otarx)nida +A 9 for each x. . * The v specifies the allocation of labor to firms. Eq ion (28) asserts that labor supply equals demand. The function y(-.x) describes the allocation of income to individuals in each state. The constraints (29) Fequire that, in each state x, the supply of the commodity equals demand ‘The Pareto-optimal allocations can be studied by introducing arbi- y linear social welfare functions. Specifically, let A be an arbitrary ® Notice that the notation embodies the assumption thatthe ouput ofall firms i alfeced by the same random variable &. Specifically, "slate of nature” is complete Aetined by x, the value taken by. If diferent firms were alfected by different random ‘arable, the description of “tate” woul have to specly the vale taken by each of these vatabls736 JOURNAL OF POLITICAL ECONOMY. Lebesgue measurable fu social welfare function is tion A:[0,1] > [0,1]. The corresponding { * Naw (y(a, 2).0) (da). (30) IFT, A,» and y(-,-) are chosen to maximize (30) subject to the constraints (28) and (29), the result is a Pareto-optimal allocation. In order to describe the Pareto optimal allocations, we study the so- lutions to these maximization problems for arbitrary \ funetions, First notice that it is possible for a planner who wishes to maximize (80) subject to the constraints (28) and (29) to ignore the identity of individuals who become workers and entrepreneurs and concern himself only with the number of entrepreneurs, that is, the number of firms. A similar simplification is possible in choosing ¥; only the distribution of labor to firms matters; itis unimportant which entre- preneur runs which firm, This makes it possible to establish a conver tion that facilitates the comparison of efficient allocations. with equilibrium allocations. Specifically, we ean assume that in making his choice of [ and A, the planner simply chooses an individual & (&-can Iso be interpreted as the number of firms) and then assigns & = [0,4] nd P= (4,1 The second simplification which is possible in the discussion of un- constrained optimality is introduced because g exhibits decreasing. returns to scale, Under this assumption, efficiency requites that every firm produce the same amount. If this were not true, output in each statex could be increased if labor were ansferred from a high output firm to a low output firm, This transfer would increase output be- cause of the differences in labor's marginal productivity (in every x) which would result from the initial inequality of the outputs of the two firms considered." Since y(a) must be the same for all entrepreneurs (28) reduces to ep for all € [0,4]. Using this result, (29) becomes | 'ylaxruday = ag“, |+a (32) "This result can be derived immediately by writing the Euler equation corresponding to the maximization with respect to). We get ED, (e3) = 8 where By i the ‘nultiplier associated with (28) and d(¢} are the mulipiers aswciated with (29). This Eb(sitola)s) = EG)guola')s) for ala = aq’ = a. Since gy isa decreasing function of E for each s,o(a') # wla}would make this equality impossible,“THEORY OF FIRM FORMATION 737 The program for obtaining Pareto-optimal allocations reduces to as, | “nla) Ew (xla.i).a)u(da) subject to (82) for allx, The firstorder conditions are Na)m(x)ur(y(a.x).2) = 8(x), forall a and allx (33) [foe fc iL where (x) is the multiplier associated with the resource constraint in x(x) is the value of the objective probability clensity ion at x. Using the value of (x) defined by obt: 8) and inserting in (84) we ‘alter taking the expectation over x, Ae ulsed.a)e( +54 a= Eulsastha)e( +a) (35) for every a Using (88) for wo different x values, say x, and x, and for two different « values, say a and , we also obt u(y(a.x,), a) _ wi(9(B.x,), B) w@x) a) wi(y(B.x).B) for alls and all a, 6 (36) be viewed as that which determines the efficient, that is, the optimal division of individuals between workers and preneurs, In the special case where (16) holds, that is, when there are stochastic constant returns to scale, (35) reduces to Caveat] om which is the 4 level which maximizes the output ah ((1 — &¥@)x for The conditions (36) are those which characterize efficient alloca- (35) and (36) that there will be several sources of inefficiency in an738 JOURNAL OF POLITICAL ECONOMY who have varying attitudes toward risk. For that reason different entrepreneurs will produce different outputs. This is one source of nefficiency. Note, however, that it will al to arse if all entrepreneurs hhave the same utility function and therefore the same attitude toward risk, A second type of inefficiency arises because of the fact that only ‘entrepreneurs bear risks in equilibrium. This is the institutional con- straint on the allocation of risk bearing of which we spoke earlier. n-general, the conditions (36) cannot be satisfied if there are risk averse entrepreneurs. A special case in which this problem does not arise occurs when all entrepreneurs are indifferent to risk. In that case condition (36) holds in equilibrium because of the linearity of entrepreneurs’ utility funetions and the Fact that workers beat We will return to discuss this case more completely later The third source of inefficiency which requires more discussion is, the optimal choice of é. To discuss this problem in an approp) setting it seems necessary to consider an equilibrium in whic entrepreneurs are the same and produce the same output. ‘This climinates inefficiencies of the first type mentioned and makes it possible to ask if (35) might be satisfied To study this question, recall that (17) is the necessary condition for entrepreneurial expected utility maximization. In general, (17) dif= fers from (85) because, as we shall see below, AkujA + gl (38) (In an equilibrium in which all entrepreneurs have the same utility function, L(w(a), @) = 124 since supply equals demand.) Also recall that, in equilibrium, w satisfies (18) where L. Thus w is the certainty equivalent of the random variable g¢ 1-4 = w tS). When entrepreneurs are risk averse, condition (18) implies that 1-4, x)-w (39) (40)THEORY OF FIRM FORMATION 739 Substituting (40) in (17) 1 Bu(A + g(L,3) ~ wh, gulls sEuj(A + g(L,%) ~ wh, &) LEg(L. aE A + g(L,3) ~ wh, a) ~ pu (A + gil.) — wh, 4) ah) where L. Risk aversion (1, < 0) also implies that > 6 = covlgi¥).u(A + g(L.8) ~ wh. 4)) Pg(L,3)u(A + giL,%) wl, @) (42) Fg(L,2)MEu(A + g(L..8) ~ wh, A) (1 = @/é. Combining (41) and (42) we obtain (A + g(L,%) ~ wh, Agel, 8) = Egil, X)u(A + g(L,5) = wh, &) ). (43) =e = plu,(A + g(L,3) — wl, with L = (1 ~ @f@. Note that (43) and (35) are the same if the covariance ¢ equals the negative of pui(A + g(+ In this case the equilibrium is efficient. Otherwise, risk aversion causes two types of errors. One of these, measured by the term — pul + =a 4 ) is introduced by the “entry condition” (18) and tends to cause the righthand side (RHS) of (35) to exceed the left-hand side (LHS). The other type of error is measured by c. It enters through the entrepreneurial maximization condition (17) ittends to make the RHS of (35) smaller than the LHS in equilibrium, To identify the direction of the effect which each of these errors has oon the choice of & consider the case in which (16) holds so that the optimal choice of é is independent of the preferences and the income Aistribution. Recall that in this ease é should be chosen to maximize output in each state and that (35) reduces to the first-order condition for this output maximization problem. It is also easy to verily that when the LHS of (35) exceeds the RHS in equilibrium, then the derivative, — 1)x — La'(4 — 1), of output with respect to ais negative at the equilibrium. It is then clear from figure 5 that in this case the equilibrium & is too large to be efficient, Thus there are to many firms in equilibrium when the error, ~¢, introduced by the0 JOURNAL OF POLITICAL ECONOMY 7 i i { ' ' ' ' Efficient al Equilibrium @ op-165, too few entrepreneurs when the error introduced by the entry condition outweighs the error introduced by the entrepreneurial first-order condition, These con- clusions coincide with intuition, On the one hand risk aversion should cause 100 few individuals to become entrepreneurs and this should operate through the entry condition, On the other hand risk aversion fon the part of those who become entrepreneurs reduces labor d mand when (16) holds (recall theorem 3). The error caused by e treprencurial risk aversion should reduce the equilibrium k mand and the equilibrium wage. ‘The low wage cre: for too many individuals to become entrepreneurs. We can now consider several important special cases. The first such, case is that in which all entrepreneurs have const aversion in the sense of Arrow-Pratt, that is, u(T,aq) r > 0; and the production function is g(b.x) = Ls (44) where y € (0,1). In this case, (35) reduces to a@=1-y (45) and the efficient labor input per firm is“THEORY OF FIRM FORMATION m4 The equilibrium conditions (17) and (18) combine to yield ke 1 we tH) 0= Sar * Gye Ty 8 BY a7 where L = (1 ~ aia I-can be shown that the RHS of (47) is a decreasing function of L and that itis negative if y/(1 — y) is substituted for L. Thus the efficient L level exceeds the equilibrium level, As a result there are too many firms in equi Note that in this class of examples the efficient number of approaches zero ify approaches one, that turns to scale become constant. The limiting case in which returns to scale indeed constant is one in which, in general, there are too m: entrepreneurs. The efficiency analysis just carried out does to this case because it assumes that gu, = 0 for all. (The ex equilibrium proof does apply if wi, < 0.) Its possible to analyze this, case directly however by substituting (Lx) = ML x in (34). The output in state x then becomes k(1 ~ @x which is clearly maximized for each x when @ = 1. Thus the measure of the optimal set of entrepreneurs is zero and almost all individuals should work, This result occurs because the technology set of the economy is the same if there is either one entrepre rger entrepreneur set of measure zero, oF 3 set of positive measure. There is one important case in which the equilibrium is efficient That is the case already mentioned in which all entrepreneurs are indifferent to risk. Since the preceding discussion makes it clear that entrepreneurial risk aversion causes the errors which result in a nonoptimal equilibrium level for é it should not now be surprising that the equilibrium is efficient when entrepreneurs are not risk averse, For that case, we have already noted that the distribution of risk is efficient, that is, condition (36) is satisfied. The linearity of the ¥y function implies then that u(",a) is independent of y(ax), s0 (85) reduces to Lx (48) a9) (50) isk indifference implies that p = 0 so that (40) becomes eS } on742 JOURNAL OF POLITICAL ECONOMY Taken together the equilibri and the equilibrium therefore The preceding discussion of efficiency suggests that entreprenew ial risk aversion is the source of all of the observed inefficiencies. When entrepreneurs are risk averse the equilibrium is not only characterized by an inefficient distribution of risk; there will, in ge eral, also be too many or too few firms and they will not employ the correct number of workers. In fact, itis well known that distribution of risk is inevitable with any equilibrium, subgroup of risk averse individuals (in this case, the entrepreneurs) bear all of the risks. Suppose, however, that we concede the inevita- bi bution of risks and ask if, given this kind of inefficiency, the other aspects of the equilibrium might be efficient Specifically, let us accept the fact that entrepreneurial risks cannot be shared and require only an optimal division of individuals between entrepreneurial activities and labor and of labor between firms. Is it then possible that in this restricted sense the equilibrium is efficient? In order to pose this question formally we define a restricted feasi- ble allocation to be a specification of F, A together with two functions, p:d-> [0.) and £:[0,1] —> [-A,2) which satisfy the equations (28), m equations (50) and (51) imply (49) ishes all of the conditions for ef |' eae =0 62) and A + g(v(a)x) + (a) = 0 for all x and a € A. ‘The function (a) specifies the labor to be employed by entre- preneur a, that is, for each ae A, »(«) is e’s labor demand. Equation (28) expresses the equality of labor supply and demand. The function € specifies the amount of ain payment made to each a. Equation (52) is a supply-cdemand equality for these payments. It guarant that the resources exist (0 make all payments. The final cor rules out bankruptcy The difference between this concept of restricted feasibility and the wotion of unrestricted feasibility should be noted. In the def ed feasibility, the payments made to individuals are contin= he state of nature x, As in an Arrow-Debreu economy, the nt on the allocation of contingent claims is that supply ‘sks can be completely reallocated, In contrast, a restricted feasible allocation specifies, for each a, a payment g(a) which is not state contingent; it does not permit a reallocation of risks. Asa result, the distribution of risks implied by a restricted feasible allocation has two important features. First, as in equilibrium, workers bear no risks; entrepreneurs bear all risks, In addition, the distribution of risk among entrepr on ofTHEORY OF FIRM FORMATION 743 determined by the distribution of labor since, for entreprenet y(a.x) is restricted to equal A + g(v(a).x) + Ela). This feature is shared with equilibrium allocations of risk since, in equilibrium, y(«,x) = g(L(u,a),x) ~ wh(w,@) + 4 if a isan entrepreneur. These consid- erations permit us t0 observe that the conditions defining restricted feasibility do indeed embody the institutional constraints on risk trad ing implicit in our equilibrium concept. It should also be noted that this definition of restricted feasible allocations does not employ any explicit or implicit assumptions about the independence or dependence of the #’s which enter the produc tion functions of different firms. An allocation A which is restricted feasible is said to be restricted efficient if there is no other restricted feasible allocation A* which Pareto dominates A. We can now prove that an equilibrium is ef sense just defined. It should first be en theorem we can and will drop the assumption that the & is the same for all firms. In fact, itis not necessary for the theorem to make any assumptions about the dependence or independence of the random production functions? THEOREM 6.—An equilibrium is restricted efficient ‘ent in the restricted. PROOF —In an equilibrium allocation a= (0.4) P=) He) = Ln) aa 63) a jel w@.e) teed se) = Ny ifaed Now consider some other allocation A* 9 dominates the equilibrium. To express this domin AD, A*, ve, €) which, ion for- Pa ally it is necessary to first ps sets i) those in A*N A, ii) those in A* OT, iii) those in P* 9, and iy) those in P* 98. If @ is in A* 9 A, then Pareto dominance of \* implies that Fuld + g(L(w(@), @),8) = w(@L (wl), «), 0) ee SEu(A + g(v*(a).3) + &(@), a), * Note that unlike the definition of unrestricted feasibility the definition of estcted feasiity emnbudies no implicit assumption about the dependence of the fs faced by different firmsm4 JOURNAL OF POLITICAL ECONOMY: n of L(w(@), @) then (a) =~ w(@W* (@). (55) By del If ais in A* 1 P, then again Pareto dominance of \* implies Eu(A + g(L(w(@), @),%) = wh (w(@), a), a) u(A +w(@), a) 6) = Fu(A + g(a), &) + Ea), a) and again (55) holds. Mais in PAP, then u(A +w(@), a) Su(A + Ea), a) 67) nd w(a) = E*@), (58) Finally, if @ is in P* 9 A, then (A + wd), a) SEn(A + g(L(w(@), @),% = ulg(a), a) so that (58) holds in this ease also. 58) holds, In fact, a similar Pareto dominates the equilibrium, there must either be a A* subset of positive measure on which (55) holds with a striet inequality or a P* subset of positive measure on which (58) holds with a strict inequality Thus [eerie > [=] eee + nat) fot (60) Inequality (60) implies that A* cannot be restricted feasible. Specifically, because of (60), equations (28) and (52) cannot hold jultaneously, We have therefore shown that the equilibri icted efficient because it cannot be Pareto dominated by a re- stricted feasible allocation. | | Restricted efficiency is similar in spirit to Diamond’s (1967) con- ied efficiency, that is, efficiency under the given constraints on the risk allocation. Here we have imposed more constraints than in Diamond's model of the stock market since, in our approach, entre~ prencurs are not allowed to share any risk with workers or other entrepreneurs. However, we get restricted efficiency without any technological assumption such as those imposed by Diamonds as- sumption of stochastic constant returns to scale. A similar result is obtained in Kihlstrom and Laffont (1978), where risk sharing is in-THEORY OF FIRM FORMATION 745 troduced by the existence of markets for shares to firms. The result- ing equilibrium is shown to be efficient in the sense of Diamond. The efficiency theorem involves no restrictive assumptions about technol- ogy. Specifically, it does not require stochastic constant returns to, scale. VII. Summary In this paper we have introduced a simple general equilibrium model of firm formation in which production requires entrepreneurial well as normal labor inputs. Workers receive fixed wages while trepreneurs receive risky profits, Individuals decide whether to b \repreneurs or workers by comparing the risky returns of neurship with the nonrisky wage determined in the compe! The wage adjusts to the point where the supply of workers is equal to the entrepreneurial demand for labor Although we have not discussed the interpretation of the entre- prene Fibution to the productive process we have implicitly or explicitly made assumptions about the nature of this contribution. The primary assumption is that the relationship between output and. the entrepreneurial labor input is characterized by an indivisibility Specifically, each firm requires a unit of entrepreneurial labor re- gardless of how much normal labor it employs and how much produces. In this sense, the expenditure of entrepreneurial labor ca be viewed asa set-up cost. Normally, the nonconvexities introduced by indivisiblities in general and set-up costs in particular cause problems when the existence of equilibrium is studied. In our model, this problem is avoided, as it can be in general (see, e.g., Aumann 1966), by assuming that the set of individuals is a continuum, ‘One possible interpretation of our model is that the entreprene contributes managerial and organizational skills. In Knight's words he performs the “function of exercising responsible control.” In fact, ire model can be viewed as a formalization, for a special case, of Knight’s discussion of the entrepreneur In our mode treprencur is characterized by two activities. He supplies entre- preneurial inputs and bears the risks associated with production, In * Koighe (1920), p. 278, » Knight’ view of the enteepreneur as well asthe view formalized here are rather different from that set forth by Schumpeter (1434, 1839), Schumpeter viewed the ‘entrepreneur as an innovator. Bee, ¢, the discussion on pp. 132-86 of Schumpeter [1934]) His view of the entrepreneur’ contribution and of his compensation i fessence dymamicy Me also specifically asserts on p. 187 of Schumpeter (1994) that “the ‘entreprencur is never a rk bearer.” For a more modern diseussion of the entre preneur and his role (or lack of one) in economic theory, sce Baumol (1968),748 JOURNAL OF POLITICAL ECONOMY Knight's treatment the entrepreneur makes the same contributions “with the performance of his peculiar twofold function of (a) exe: ising responsible control and (b) securing the owners of productive services against uncertainty and fluctuation in their incomes." Knight's view of the labor market and of an individual's decision to become a worker or an entrepreneur also appears to be similar to that formalized here. This is illustrated by the discussion on pages 273-74 of Knight (1921). Specifically, he asserts that, "the laborer asks wh he thinks the entrepreneur will be able to pay, and in any case will not accept less than he can get from some other entrepreneur, or by ‘turning entrepreneur himself. In the same way the entrepreneur offers to any laborer what he thinks he must in order to secure his services, and in any case not more than he thinks the laborer will be worth to him, keeping in mind what he can get by turning laborer himself" He continues: ‘Since in a free market there can be but one price on any commodity, fa general wage rate must result from this competitive bidding." Our model represents only a special case of Knight’ view because ‘we assume that all individuals are equal in their ability to perform al as well as normal labor functions. They differ only in ngness to bear risks. Knight emphasizes ability as well as ness [and] power to give satisfactory guarantees” as factors determining the supply of entrepreneurs. In our model, the size of the initial income A can be interpreted as a measure of an individual's power to guarantee wages by bearing risk. We have assumed he dividuals are alike in their possession of this “power.” An in- ng alternative interpretation can be made by explaining the differences in risk aversion as arising from differences in wealth Suppose, for example, entrepreneurs have the same utility function which is decreasingly risk averse (in the absolute sense of ). Then the differences in the willingness to bear risk— that is, to “give satisfactory guarantees” in Knight’s words—will be determined by initial wealth. Thus if A varies across individuals, the more risk dividuals will also be those who initially are the poorest. Assuming that the constraints = Alu are never binding in equilibrium our model can then be reinterpreted as predicting that entrepreneurs are those who are initially wealthy. (OF course the opposite prediction can be obtained by making the less generally accepted assumption that the common utility function is increasingly risk averse.) With this interpretation, the possession of wealth which that © Knight (1921, p. 278) "Ibid, pp, 273-74 (emphasis added. Thi pe 2 Thi 288,THEORY OF FIRM FORMATION a7 provides additional power to give satisfactory guarantees also makes, an individual more willing to bear risk To complete the analogy between our results and Knight’s discus- sion note that our theorem 4, which relates the equilibrium wage t© the level of entrepreneurial risk aversion, was in a sense anticipated by the discussion of Knight (1921, p. 283) which concludes that “entre- preneur income, being residual, is determined by the demand for these other [productive] services, which demand is a matter of self confidence of entrepreneurs as a class..." This paper has extended the classical results concer tence and stability of equilibrium to the entrepreneurial model. We ce also described the nature of the equilibrium’s inefficiencies and identified the institutional constraints on risk trading as the source of these inefficiencies. These results establish that it is possible 10 con- struct an internally consistent general equilibrium model of entre- preneurially operated firms. In fact, this analysis should only be viewed. as a first step in the construction of a general equilibrium entre preneurial theory of the firm under uncertainty. As presented here itis, pethaps best viewed as a description of equilibrium in a world of sn businesses or farms. The next step is to incorporate a stock market into our analysis. Once a stock market is embedded in the model, we can use it to ask ‘cresting questions about the interaction between a modern firm's financial and productive decisions. Furthermore, the introduction of a stock market represents an institutional change that facilitates risk trading and thereby eliminates some of the inefficiencies that occur at aan equilibrium. The extension of the model in this direction has be stuclied in a subsequent paper (Kihlstrom and Latfont 1978) References Arrow, Kenneth J. “The Role of Securities in the Optimal Allocation of Risk: Bearing.” Rev. Econ. Studies 81 (April 1964): 91-96, Essays in the Theory of Risk Bearing. Chicago: Markham, 1971 Aumann, Robert J. “Existence of Competitive Equilibria in Markets with a Continuum of Traders.” Econometrica 34 (January 1966): 1=17 Baron, David P, "Price Uncertainty, Uility and Industry Equilibriv Competition.” Internat. Econ, Rev. 1 (October 1970): 463-80, Baumol, Wiliam J. “Entrepreneurship in Economic Theory.” A.E.R. Papers ‘and Proc. 63 (May 1968): 64-71 Brock, William A. "On Models of Expectations Th Behavior of Economic Agents over Time.” J. Econ. Theory 1972): 348-76. n Pure Arise from Maximizing (December abi748 JOURNAL OF POLITICAL ECONOMY Debreu, Gerard. Theory of Value. New York: Wiley, 1959. Diamond, Peter A." The Role of a Stock Market in a G Model with Technological Uncertainty.” AER, 759-16. Deze Jacques, “Investment under Private Ownership: Optimal, Eau ium: Stability.” In Allocation vender Uncertainy: Equilibvium and Optimality, London: Macmillan, 1974 Kihlstrom, Richard E., and Laffont, Jean-Jacques. “A pprencurial Model of the Stock Mar ois, 1978, Frank H. Risk, Uncertainty and Profi. New York: Harper & Row, neral Equilibrium September 1967) mpetitive: Entre- 1” Mimeographed. Urbana, IL: Univ 19 Pratt, John W. “Risk Aversion in the Small and in the Large.” Feonometrica 32 (atiuary-April 1964): 122-86, Quandt, Richard, and Howrey, E. Phillip. “The Dynamics of the Number of Firms in an Industry.” Rev. Eeon. Stulies 35 (July 1968): 349-53. Radner, Roy. “Competitive Equilibrium under Uncertainty.” Econometrica 36 anuary 1968): 31-58, Schumpeter, Joseph A. The Theory of Economic Development; An Inguiry into Profits, Capital, Credit, Interest, and the Busines Cycle. Cambridge, Mass. Harvard Univ, Press, 1934 Business Cycles; A Theoretical, Historical, and Statistical Analysis of the Capitalist Process. Abridged ed. New York: McGraw Hill, 1939. smith, Vernon, “Optimal Costly Firm Entry in General Equilibrium.” J. Econ. Theory 9 (1974): 397-417, Symposium on the Optimality of Competitive Ca Management Sci. 5 (Spring 1974): 125-86, al Markets." Bell J. Bon.