The Entrepreneurial Critique of the Optimization Paradigm

ROBERT F. MULLIGAN

mulligan@wcu.edu

828-227-3329

Associate Professor of Economics, Department of Business Computer Information Systems & Economics, College of Business, Western Carolina University, Cullowhee NC 28723; Research Associate, Department of Economics, State University of New York at Binghamton

Abstract. The optimization paradigm in neoclassical economics is defined and criticized. Although the optimization paradigm provides the foundation for most mathematical formalism in modern technical economics, the paradigm rest in turn on extreme and extremely unrealistic information assumptions. No scope for entrepreneurial activity is admitted. This paper argues that entrepreneurial change is one of the most basic ingredients for understanding the functioning of markets and the coordination of economic activity. Several extensions of the standard model are presented and discussed. Alternative views of market equilibrium offered by Schumpeter and Kirzner are developed, and the relationship between the two competing views of equilibrium is examined.

Key words: optimization paradigm, entrepreneur, resource allocation, productive activity

JEL classification: B41, C61, C62, D11, D40.

"What we cannot speak about we must pass over in silence."

Ludwig Wittgenstein[1]

This paper criticizes the optimization paradigm in neoclassical microeconomics with respect to its treatment of entrepreneurship. Conventional cost minimization-profit maximization exercises central to modern neoclassical economics virtually ignore entrepreneurial action. Entrepreneurial managers should be understood as seeking to (a) adjust the production process to extract more output from each input or more output from the same level of inputs, (b) switch inputs or combine new inputs with those already used to produce more or cheaper output, (c) devise strategies to sell their output or some part of it for a higher price, (d) attract new buyers and devise new uses for the output, and (e) lower the costs of their inputs. In the theory of the consumer, the optimization exercise ignores entrepreneurial innovations which imply changes in the (a) number of utility function arguments, (b) form of the utility function, (c) consumer income, (d) prices of consumed goods, and (e) quality of consumed goods. More fundamentally, the optimization paradigm ignores entrepreneurial consumers' alertness to these opportunities to costlessly improve the satisfaction of their wants.

In addition to developing an Austrian critique of the prevailing orthodoxy, this paper also offers some modest extensions which partially address the critique, though they fail to overcome it. It is hoped that as the optimization paradigm continues to be central to what is taught in graduate economics programs and much economic research, its practitioners will come to accept a more mature understanding of its limitations.

1. Introduction

The optimization paradigm in neoclassical microeconomics is the modeling of economic behavior as a series of constrained and unconstrained optimization problems. Market participants are thought of as knowingly optimizing known objective functions: minimizing cost, maximizing output, maximizing profit, or maximizing utility, subject to known constraints (Robbins 1935). However, analytical functions can never be more than imperfect analogues of the processes of producing output or obtaining satisfaction through consumption[2]. In a sense, considered as mathematical models, they do not purport to be anything more. The optimization paradigm is based on fundamental assumptions precluding the possibility of entrepreneurial activity. The uncertainty entrepreneurs overcome and exploit to earn entrepreneurial profits is removed by assumption.

The optimization paradigm is usually presented through graphical analysis at the undergraduate level, while graduate courses are typically more formalized (Samuleson 1947, Debreu 1959, Silberberg 1978, Henderson and Quandt 1980, Chiang 1984, Varian 1984, Takayama 1985). The optimization paradigm assumes away the entrepreneur and treats all the data of the market as perfectly known in advance. This approach fundamentally misrepresents the nature of market competition. The Austrian school has always been skeptical of mathematical formalism: "in the imaginary construction of an evenly rotating system nobody is an entrepreneur and speculator. In any real and living economy every actor is always an entrepreneur and speculator...." (Mises 1949: 252). Entrepreneurial action sometimes improves market coordination and moves the market toward a hypothetical equilibrium, but also sometimes increases the level of discoordination, moving the market further away from idealized equilibrium. Thus entrepreneurship cannot be defined exclusively in terms of its relation to either equilibria or coordination among economic planners.

<<Table 1 about here.>>

Entrepreneurial action is necessarily multifaceted and defies simple quantification. Blaug (1998: 227) cites several different historical views of entrepreneurship. Entrepreneurial action includes arbitrage (Cantillon 1755), coordination (Say 1803; Kirzner 1973), innovation (Schumpeter 1934), uncertainty-bearing (Knight 1921), and most recently (Casson 1982, 1985) increasing the range of available judgments on resource allocation. There is no simple way to incorporate all modes of entrepreneurship in an optimization problem. Entrepreneurs change objective and constraint functions, subverting the analytic character of the optimization paradigm.

The basic problem with the optimization paradigm may be formulated as an unwarranted assumption that all information summarized in market prices, including the preferences of others, is known to agents in advance of their decision. Thus the optimization model of economic decision-making is perfectly deterministic. However prices cannot be known with certainty until after an exchange occurs. In fact entrepreneurs do not optimize with respect to any objectively knowable information set.

A more meaningful way to consider entrepreneurial behavior is that it proceeds experimentally and so generates the information of the market, which can never be known in advance of a voluntary exchange, but only after one takes place (Buchanan 1986), and only as having no necessary implications for potential subsequent exchanges. Once an individual has bought or sold a good or service at a certain price, special to the time, place, and parties to the transaction, unalterable historical information of the market has been created. But this information does not necessarily have any significance for the future, and thus objective knowledge cannot form the basis for market participants' optimizing behavior.

Real behavior is better captured through less precise "rules of thumb" than through the supposedly rigorous optimization paradigm. It seems clear to even casual observers that human behavior is not based on mathematical optimization, either underlying or explicit. Because much human knowledge is tacit, and thus inherently decentralized and subjective, it becomes particularly difficult to justify assumptions of perfect knowledge and foresight which underlie the optimization paradigm.

The rest of this paper is organized as follows. Section two discusses the entrepreneur emphasizing the Austrian tradition. Section three presents a discussion of cost minimization. Section four presents a parallel discussion of profit maximization. Section five introduces a theory of the entrepreneurial consumer. Section six discusses the distinction between Schumpeter's and Kirzner's definitions of entrepreneurial action. Finally, section seven presents concluding remarks.

2. The Entrepreneur's Function in Society

Schumpeter (1934) identifies five types of entrepreneurial innovation (1) introducing new outputs or improving the quality of existing outputs, (2) introducing new methods of production, (3) opening new output markets, especially new export markets, (4) finding new sources of supply of raw materials or intermediate inputs, and (5) creating new kinds of industrial organizations. To Schumpeter, entrepreneurs are not inventors but decision makers who allocate resources to exploit inventions; they are not risk-bearing capitalists but borrow funds from the capitalists to finance their innovations. Casson (1987: 151) notes that Schumpeterian entrepreneurs are defined by their managerial or decision-making role.

Mises (1949: 253) defines the entrepreneur as "acting man exclusively seen from the aspect of the uncertainty inherent in every action." The entrepreneur is not merely a particular human archetype, but a category of catallactic action (ibid. 251). To Mises uncertainty is everywhere and consequently, everyone is an entrepreneur. But in the optimization paradigm, uncertainty has been assumed away along with the entrepreneur. Entrepreneurial action always responds to an uncertain future, facing uncertainties of future consumer demand, changes in demand and supply in factor markets, and differences in the ability of other entrepreneurs to foresee with a compatible vision, the prerequisite for mutually compatible plans (Mises 1949: 212-214). The entrepreneur always faces uncertainty because production takes time, so any entrepreneurial action necessarily involves "speculation in the anticipation of future events" (Greaves 1974: 39). Lachmann (1986: 65-70, 116-117, ) also emphasizes that entrepreneurial action responds to uncertainty about the future.

Much of Kirzner's (1973, 1979, 1984a, 1997) work on entrepreneurship can be reformulated as a detailed critique of the optimization paradigm. Kirzner and Hayek stress the problem of dispersed knowledge given to no one in its entirety (Hayek 1949: 77-78), a problem whose "extent and seriousness cannot be known in advance," "arising out of unawareness of one's ignorance" (Kirzner 1984b: 162). Mises emphasizes the entrepreneurial response to, and exploitation of, market information: "entrepreneur means acting man in regard to the changes occurring in the data of the market" (Mises 1949: 254). The Misesian entrepreneur is alert to arbitrage opportunities which occur when prices in resource markets are not adjusted to prices in the product markets (Kirzner 1973: 85). This is a special case of Hayek's (1949) problem of dispersed knowledge.

In contrast, in some areas Hayek's (1973: 27) understanding of the entrepreneur is particularly sterile and defective[3]. In his view, business firms are designed artifacts like radios, toasters, or telescopes. Once the laws of economics are put in place, entrepreneurs exploit them to construct business organizations much the same way engineers employ physical laws to construct machines. Hayek's understanding of firm organization proceeds according to a military model, where the manager issues orders that are followed without question or exception. Hayek recognizes that the entrepreneur is endogenous to the firm, but he fails to perceive that once a business enterprise is set up, its employees enjoy some degree of autonomy, and the firm in effect takes on a life of its own.

3. The Entrepreneurial Producer I: Cost Minimization

In the optimization paradigm, the theory of the firm provides dual and equivalent cost minimization and profit maximization problems. Managers exploit a given and static production technology and a given and static vector of resource and output prices to transform a set of given inputs into a given output. Furthermore, the market structure is also given, when in fact it results from the action of entrepreneurial innovators. Entrepreneurial action ignored by the optimization paradigm includes seeking to (a) adjust the production process to extract more output from each input or more output from the same level of inputs, (b) switch inputs or combine new inputs with those already used to produce more or cheaper output, (c) devise strategies to sell their output or some part of it for a higher price, (d) attract new buyers and devise new uses for the output, and (e) lower the costs of their inputs.

It might be argued that theories of market segmentation and product differentiation allow for item (c), but this is only possible if the market is structured in a way that prevents firms from facing perfectly elastic demand. In theory, market structure is predetermined by assumption; in reality it results from the behavior of market participants. In reformulating the Walrasian theory of perfect competition, Makowski and Ostroy (2001: 480) recognize "(1) prices are not exogenously given, they arise from bargaining," and "(2) the set of active markets is not exogenously given, it results from innovation." Since the conventional view is that entrepreneurs work primarily through firms, the neoclassical theory of the firm's inability to accommodate entrepreneurial action is a major shortcoming.

3.1 The Standard Cost Minimization Problem

One way to formulate the firm's conventional optimization problem is to maximize output subject to a fixed cost constraint. The objective function for optimization is the production function relating the inputs x₁ and x₂ to the quantity of output q, and is given as

q = f(x₁, x₂)

though, unlike the consumer's utility function, the production function always has a cardinally-measurable physical or value output. Nevertheless, it remains debatable whether the production process can ever be meaningfully captured by an analytic function purporting to be, at best, an approximation. The possibility of using additional inputs x₃ through x_n is ignored. The key Austrian insight that production occurs over time (Menger 1871: 152[4]; Hayek 1931, 1933, 1935, 1939, 1941) is ignored. Non-Austrian economists are hardly ignorant of this fact, but theorize as if they were.

The constraint function is the producer's cost constraint which is conventionally given as

C⁰ = r₁x₁ + r₂x₂ + b

where r₁ and r₂ are the resource prices of input resources X₁ and X₂, b is the cost of fixed inputs, and C⁰ is the fixed total cost. Again, the cost constraint is assumed given and known in advance. No scope for entrepreneurial discovery is permitted (Kirzner 1984b: 154). Mises notes

in the imaginary construction of the evenly rotating economy there is no room left for entrepreneurial activity, because this construction eliminates any change of data that could affect prices. As soon as one abandons this assumption of rigidity of data, one finds that action must needs be affected by every change in the data. As action necessarily is directed toward influencing a future state of affairs, even if sometimes only the immediate future of the next instant, it is affected by every incorrectly anticipated change in the data occurring in the period of time between its beginning and the end of the period for which it aimed to provide (period of provision). Thus the outcome of action is always uncertain. Action is always speculation. (1949: 252)

The contrasting view of neoclassical microeconomics is that action is always optimization, and that the "correct" outcome is always predetermined by the unalterable data of the market. The optimization paradigm fails to recognize the scope for uncertainty, ignorance, and risk. The cost constraint also suffers from omission of the same arguments x₃ through x_n as the production function, as well as their prices r₃ through r_n. One means for exercising entrepreneurial action is in obtaining lower input prices[5] r₁ and r₂, or r₁ through r_n, or a combination of different inputs that minimizes total cost (Schumpeter 1934: 133-135). Entrepreneurial alertness might also be exercised in seeking to reduce the "fixed" cost b. Since b is only fixed by construction, alert entrepreneurs should seek ways of reducing it whenever possible. The producer, having chosen C⁰ or had it imposed externally, seeks to produce as much output q as possible.

The constrained output maximization problem is constructed as

V = f(x₁, x₂) + μ(C⁰ - r₁x₁ - r₂x₂ - b)

where μ ≠ 0 is an undetermined Lagrange multiplier. Finding the first partial derivatives of V with respect to x₁, x₂, and μ, and setting them equal to zero:

∂ V	=	f1 – μ r1 = 0
∂ x1
∂ V	=	f2 – μ r2 = 0
∂ x2
∂ V	=	C^0 - r1x1 - r2x2 - b = 0
∂ μ

Algebraic manipulation yields the familiar relation[6]

f1	=	r1
f2		r2

which states that, at a cost minimum, the ratio of the marginal products of resource inputs X₁ and X₂ must be equal to the ratio between their prices. The minimum found through optimization is subject to the constraint that the quantities of other inputs X₃ through X_n have been arbitrarily set at zero. These ratios between the marginal products and the resource prices define the rate of technical substitution (RTS) between X₁ and X₂. The first-order conditions can also be written as

μ =	f1	=	f2
	r1		r2

indicating the contribution to output of the last dollar spent on each input must equal the Lagrange multiplier μ, which is thus the first derivative of output with respect to cost C with prices r₁ and r₂ constant and resource quantities x₁ and x₂ allowed to vary.

The cost minimization problem can be made more realistic by allowing total cost to vary. When C is allowed to vary, the differential of the cost equation is

dC = r₁dx₁ + r₂dx₂

The fixed cost b drops out, though if entrepreneurial search aims at lowering b, it should not drop out of the first-order conditions.

The firm's problem may also be formulated as minimizing a variable cost function subject to the constraint that a prescribed quantity of output must be produced with the given technology described by the production function. The constrained cost minimization problem is expressed as

Z = r₁x₁ + r₂x₂ + b + λ[q⁰ - f(x₁, x₂)]

and the partial derivatives of Z are set equal to zero to find the first-order conditions:

∂ Z	=	r1 – λ f1 = 0
∂ x1
∂ Z	=	r2 – λ f2 = 0
∂ x2
∂ Z	=	q^0 – f(x1,x2) = 0
∂ λ

The first-order conditions allow us to solve for the following relations,

f1	=	r1	or	1	=	f1	=	f2	or	RTS =	r1
f2		r2		λ		r1		r2			r2

The Lagrange multipliers λ and μ are reciprocals. λ gives the first derivative of cost with respect to output, that is, the marginal cost[7].

3.2 Limitations of Standard Cost Minimization

The optimization paradigm is based on the principle of knowingly optimizing a known objective function subject to known constraints, removing much of the uncertainty with which real-world market participants have to contend. The standard model of perfect competition imposes extreme, and extremely unrealistic, information requirements (Makowski and Ostroy 2001: 480). Furthermore, analytical functions can never be more than imperfect analogues of the processes of producing output or obtaining satisfaction through consumption, and the Austrian school has always criticized excessive mathematical formalism. Limitations of the optimization paradigm should be recognized in that the true objective function may not be known with certainty to market participants, and may not be expressible as an analytic function, even if "known" (Kirzner 1984b: 154).

An alert entrepreneur would always seek to discover ways to adjust the production process to extract more output from each input, or equivalently, extract the same level of output from fewer inputs (Schumpeter 1934: 129-133). Instead, the optimization problem assumes a static, given technology. The optimization problem is necessarily untenable unless the technology has been optimized in advance. In reality, production technology results from the creative process of human action and entrepreneurial discovery. Technology is created, not given.[8]

(a) Static Input Set

Entrepreneurial alertness might also be applied in discovering new inputs X₃ and X₄ to either substitute or complement X₁ and X₂ in producing a given output. Because the optimization paradigm arbitrarily excludes inputs X₃ through X_n, a more realistic optimum is defined by

μ =	f1	=	f2	=	...	=	fn
	r1		r2				rn

This more global optimum still assumes a pre-determined and static technology, and can never be fully realized, as the scope for entrepreneurial alertness in discovering new inputs is inexhaustible. In neoclassical microeconomics the optimization problem is constrained by arbitrarily setting the quantities of potential inputs X₃ through X_n equal to zero. Even in the context of the optimization paradigm, imposing this indefinite number of constraints, though analytically necessary, is also necessarily unrealistic. Also this exercise, even as modified, ignores the possible choice of different outputs (Schumpeter 1934: 134).

Entrepreneurial managers may switch inputs, use new inputs, or both to produce cheaper output. This is not an optimization process. If the production technology A which is assumed uses inputs a and b, and a different production technology B uses c and d, and can produce the same output at a lower unit cost, obviously technology B should be used, but the optimization paradigm assumes this is known in advance. In fact it can only be discovered through experience, trial-and-error, entrepreneurial alertness, discovery, etc. These less tractable transactions of economic behavior cannot be modeled as optimization processes because they cannot meaningfully be represented as having predetermined outcomes.

Commitment to the more expensive, less efficient, technology A may preclude discovery of benefits potentially offered by technology B. There are always better versus worse ways to combine inputs, but never a final best, the optimum, which forever precludes discovery of improved technology. This holds even when the input set is fixed.

(b) Information Constraints

One fault of the optimization paradigm is that it assumes as given both the objective function and constraints. Provided this information is attainable, it can only be uncovered through extensive search and entrepreneurial alertness. Perhaps the most severe criticism against the optimization paradigm is that an entrepreneurial producer can carry out a valid optimization exercise based on unique, subjective, and fallible expectations of prices at which the inputs can be bought and output sold (Kirzner 1990: 167). If these expectations are not realized, the producer may lose money in spite of having optimized, because market participants often fail to optimize with respect to their true circumstances. True circumstances can only be known ex post and can only be discovered through experience, but by then it is too late. This first kind of knowledge problem (Kirzner's Knowledge Problem A) causes planned exchanges to be impossible to fulfill. Kirzner notes that these kinds of problems are self-correcting (Kirzner 1990: 169-171), as market participants either adjust their plans to recognize the realities of the market, or withdraw from the market.

It is also possible for the optimization exercise to lead to the erroneous conclusions that inputs cannot be obtained, or output sold, at sufficiently low or high prices, or that production technology, input quality, or consumer demand for the output, are actually better than anticipated by entrepreneurial planners (Kirzner 1990: 168-9). In these cases, (Kirzner's Knowledge Problem B), exchanges which are theoretically feasible, and could be seen ex post to have been feasible, are never planned or undertaken, because market participants are unaware of the feasibility of the potential exchanges. Entrepreneurs always seek to discover such opportunities, but many must go undiscovered. These kinds of problems are not self-correcting, and await entrepreneurial discovery before anyone can be aware of them. The entrepreneur can profit by uncovering, and remedying, instances of Knowledge Problem B.

These objections based on Kirzner's two knowledge problems can be given an alternative formulation, drawing mainly from Hayek (1949) and Kirzner (1984a, 1984b, 1990): the information set assumed by the optimization paradigm does not exist in reality. No person possesses it in its totality, but the optimization paradigm assumes perfect information shared by all market participants. In reality, it is more accurate to suggest that each market participant possesses some relevant information, much of which is purely subjective. Much of this information is held exclusively by a certain individual, for example, that individual's subjective preferences or his or her plans for future consumption and production. Individuals also differ in their alertness, both in terms of intensity and application (Kirzner 1979: 170). Entrepreneurs overcome the social problem of information dispersal whenever they generate flows of information that stimulate revision of uncoordinated decisions toward greater mutual coordination (Kirzner 1984a: 147) moving the market toward a never-realized equilibrium state. The very concept of market equilibrium is merely an analytical convenience with little practical relevance (Nelson and Winter 1982; Makowski and Ostroy 2001).

Prices summarize relevant information which would otherwise be useless to market participants in satisfying their wants, but the inadequacies in market prices also create the profit-and-loss incentives for entrepreneurs to adjust prices (Kirzner 1984a: 149). A price may summarize economic information regarding the supply and demand conditions in the relevant markets, without signaling whether the price represents an equilibrium or a disequilibrium. Entrepreneurs compete in adjusting prices in a "competitive process which digs out what is in fact discovered" (Kirzner 1984a: 150). The competitive process Hayek (1978: 180) describes, where "competition is valuable only because, and so far as, its results are unpredictable and on the whole different from those which anyone has, or could have, deliberately aimed at," is utterly incompatible with the optimization paradigm.

Kirzner's (1984b: 160) view is that disequilibrium prices offer pure profit opportunities for alert entrepreneurs who can arbitrage among different prices temporarily prevailing in different markets. This incentive allows entrepreneurs to effect adjustment of the disequilibrium price vector towards equilibrium. Although, by construction, an equilibrium is never reached, and if it were reached, is never persistent, the Kirznerian entrepreneur always acts to lessen the extent of the disequilibrium. The equilibrium which is approached constitutes a spontaneous order, a level of coordination which results from human action but not from human design (Hayek 1967). In contrast, a Schumpeterian entrepreneur moves the market price vector away from equilibrium, by introducing new production technologies, marketing and distribution media, and creating new plans which increase the social dispersion of knowledge. This distinction is explored more fully in section 6. Both kinds of entrepreneurship are ignored by the optimization paradigm.

The cost-minimization problem also fails to recognize the mode of entrepreneurial action which occurs when a firm incurs additional production or selling costs to make the product more desirable to the consumer (Kirzner 1973: 24). Schumpeter (1934: 135) mentions a kind of entrepreneurship consisting of a search for new markets for an existing product. In the optimization paradigm, output quantity is the decision variable, and selling price is fixed along with production technology. Increased consumer preference can (a) justify increased production or selling costs, (b) enable the firm to increase product price, and (c) be engineered by improving product quality or marketing, but these issues are ignored. This category of entrepreneurial action includes both increased production costs incurred to improve the subjective quality of the product in consumers' eyes, and increased selling costs which also aim at improving subjective quality.

Table 2 summarizes some of the limitations of the optimization paradigm and cites some instances of more realistic, less restrictive, though also less mathematically formalized, treatment of entrepreneurship given by the Austrian school.

<<Table 2 about here.>>

3.3 Extensions of Standard Cost Minimization

The production function can be adapted to demonstrate entrepreneurial innovation, but only at the cost of excessive delimitation of the choice faced by entrepreneurs.

(a) Additional Inputs

Within the optimization paradigm, additional arguments can be introduced to capture some, though not all, potential for entrepreneurial action. At least three inputs must be included, each of which can be considered a composite commodity: X₁, the bundle of inputs already used in producing output Q; X₂, the bundle of inputs subject to entrepreneurial discovery; and X₃, the bundle of inputs remaining undiscovered. The constrained cost minimization problem is then

Z = r₁x₁ + r₂x₂ + r₃x₃ + b + λ[q⁰ - f(x₁, x₂, x₃)]

and the partial derivatives of Z are set equal to zero to find the first-order conditions. The entrepreneur introduces the innovation of violating the constraint that x₂ = 0, but the constraint remains that x₃ = 0. Before entrepreneurial action, the first order conditions are:

∂ Z	=	r1 – λ f1 = 0
∂ x1
∂ Z	=	q^0 – f(x1, x2, x3) = 0
∂ λ

After entrepreneurial innovation, the first order conditions become:

∂ Z	=	r1 – λ f1 = 0
∂ x1
∂ Z	=	r2 – λ f2 = 0
∂ x2
∂ Z	=	q^0 – f(x1, x2, x3) = 0
∂ λ

Because a linear restriction has been removed, the cost minimum after entrepreneurship must always be below the before-entrepreneurship minimum, provided the constraint was initially binding. Although this optimization exercise, constructed to show the optimum before and after entrepreneurial innovation, does capture one highly delimited, highly stylized species of entrepreneurial action, many other kinds are not amenable to analysis or representation in the optimization paradigm. The kind of predefined innovation modeled here thus could not have been particularly valuable. No one learns anything terribly profound about the process of entrepreneurial innovation from conducting such augmented optimization exercises.

(b) Technological Improvement

Suppose the Cobb-Douglas production function q = f(x₁, x₂) = Cx₁^ax₂^b, represents the production technology understood by managers and engineers prior to an innovation. The prior, restricted production function has the logarithmic form

ln q = ln C + a ln x₁ + b ln x₂.

It is nested in the more general transcendental-logarithmic (translog)[9] form

ln q = ln C + a ln x₁ + b ln x₂ + c ln x₁² + d ln x₁x₂ + e ln x₂².

Suppose a particular entrepreneurial discovery consists of learning how to make use of the second-order combinations of inputs in the translog. The resulting, more flexible, optimization exercise will realize lower costs and higher profits, because restrictions on the second order terms in the production function have been lifted.[10]

The difficulty faced by the approach of modeling entrepreneurship as a removal of known constraints is that if the less restrictive technology were known to the entrepreneur in advance, they would have made full use of it. This construction can only be set up in an environment where the before and after optima are both fully determined, and only the entrepreneur is ignorant of the difference. This is a particularly sterile way of modeling entrepreneurship, because it constrains the entrepreneur exclusively to realizing artificial productivity gains only as defined in advance by the theorist. Real world entrepreneurs recognize no such constraints. Thus one way of expressing the limitations the optimization paradigm imposes on modeling entrepreneurial behavior is that it reduces entrepreneurship to an Easter egg hunt, where the only prizes which can be uncovered by entrepreneurial discovery, must always be defined in advance by the theorist. This is a reductio ad absurdum against the optimization paradigm. The real state of affairs is that entrepreneurs often uncover advances undreamed of by anyone else, or even by themselves prior to the discovery.

4. The Entrepreneurial Producer II: Profit Maximization

4.1 Standard Profit Maximization

The firm's cost minimization problem is often represented by its dual. Since the entrepreneur is free to vary cost along with output, the unconstrained profit maximization problem is written as

π = pq – C

where p is the output price, assumed to be given. Substituting in the production and cost functions, the profit function is conventionally written as

π = p f(x₁, x₂) – r₁x₁ - r₂x₂ - b

which is maximized with respect to the choice variables x₁ and x₂. Again, limitations imposed by assumptions of static technology and limited inputs invalidate the exercise. It should be emphasized that this algebraically unconstrained optimization problem yields an economically constrained solution, as a consequence of the assumptions which are imposed. The first-order conditions[11] are

∂ π	=	pf1 – r1 = 0
∂ x1
∂ π	=	pf2 – r2 = 0
∂ x2

which can be written as

pf₁ = r₁ and pf₂ = r₂

The partial derivatives of the production function with respect to the inputs are the marginal products. Multiplied by price, they provide the marginal revenue products, which are set equal to the input prices r₁ and r₂ at the profit maximum. This is the familiar condition that an input is used as long as its marginal revenue product exceeds or equals its cost to the producer. However, the constraints that the quantities of additional potential inputs X₃ through X_n are fixed at zero, guarantee this is a constrained optimum.

4.2 Limitations of Standard Profit Maximization

Because all search activity is ignored, important features of economizing action are disregarded and assumed away (Kirzner 1984: 156) For example, one forum for entrepreneurial action is to seek monopoly ownership of a resource. This broad category of human action includes both cornering the market for a particular input, and creation of any form of intellectual property. Schumpeter (1934: 152) discusses how temporary monopoly profits always accrue to innovators. Entrepreneurs pursue such strategies all the time. Because the optimization paradigm assumes input and output prices as given, this mode of entrepreneurial action is necessarily ignored. Entrepreneurs should always desire to be monopolist resource owners because that allows them to charge a higher price for their output than non-monopolists (Kirzner 1973: 21). Entrepreneurs also seek to increase output price through becoming monopoly suppliers of the output, or monopolistic-competitive suppliers, for example, through creating intellectual property in their output, or through advertising. Recognizing that market structure is determined by the competitive process of the market is part of Makowski and Ostroy's (2001) reformulation of the Walrasian general equilibrium model of perfect competition.

The price of output is assumed given, thus the vector of prices is not permitted to adjust to realize efficient allocation. Producers are not permitted to adjust the type of output, which is also assumed given. One obvious area for entrepreneurial innovation is offering different kinds of output for sale to consumers (Schumpeter 1934: 134-135). The role flexible prices and their adjustments play in overcoming Hayek's (1949) problem of dispersed knowledge and coordinating the plans of producers and consumers, is similarly ignored (Kirzner 1984a: 139-140). Kirzner notes that equilibrium prices, if they could persist, would signal market participants' plans to each other, and thus guide future planning. In contrast, disequilibrium prices signal to alert market participants how revised plans may benefit market participants in the future. In Kirzner's view, disequilibrium prices predominate, thus entrepreneurial opportunities are everywhere. Entrepreneurial consumers and producers who take advantage of these disequilibrium prices move the market toward dynamic equilibria which are generally never realized.

4.3 An Expanded Profit Maximization Exercise

The profit maximization exercise could be expanded to realize a more global optimum, which would be represented by the n conditions

pf₁ = r₁, pf₂ = r₂, ... , pf_n = r_n

but it should be emphasized that this optimum can never be realized. In reality, most of the n conditions for a global profit maximum are necessarily invalidated. Economic agents' ignorance sets all but a few of the x_is equal to zero. Thus the realized profit maximum is constrained. The solution also assumes a fixed output price p and fixed input prices r_i, and the assumption of static production technology ensures the first and second derivatives (f_is and f_ijs) of the production function are also constant. Entrepreneurial activity can be described analytically as removing constraints on p, and the x_is, r_is, f_is, and f_ijs, however, this fails to capture all entrepreneurial action, as entrepreneurs also seek to produce those outputs which will yield the highest profits.

5. The Entrepreneurial Consumer

We do not generally think of entrepreneurship outside firms. However, given Mises' definition, "acting man exclusively seen from the aspect of the uncertainty inherent in every action," (1966: 253) it is clear the consumer can be an entrepreneur in the act of consumption. In the neoclassical theory of the consumer, consumers maximize an unknown and unobservable utility function, which must be specified in a finite number of arguments, subject to a budget constraint. In a formal sense, an entrepreneurial consumer would always be (a) trying out new arguments in the utility function, as well as (b) trying out new ways of extracting additional utility from old arguments. For example, entrepreneurial consumers would experiment with new preference functions in an effort to shift their indifference curves inward. The neoclassical conception ignores both kinds of entrepreneurial behavior, by assuming they have been carried out in advance.

To account for the possibility of entrepreneurial consumption, the number of arguments in the utility function and budget constraint must be made indefinite, as with the cost and profit functions. In addition, the entrepreneurial consumer is always trying to increase his or her income, as well as searching for lower prices and higher quality arguments for consumption. Entrepreneurial consumers are alert to improving (a) the number of utility function arguments, (b) the form of the utility function, (c) consumer income, (d) the prices of consumed goods, and (e) the quality of consumed goods. More fundamentally, the optimization paradigm ignores entrepreneurial consumers' alertness to these opportunities to costlessly improve satisfaction of their wants.

5.1 Conventional Utility Maximization

Kirzner (1973: 18) notes the convention of treating consumers as merely passive price-takers is only an analytical convenience. The consumer's optimization problem is usually defined as maximizing a preference or utility function subject to a budget constraint. In the two-good case the budget constraint is

y⁰ = p₁q₁ + p₂q₂

where y⁰ is the consumer's fixed income and p₁ and p₂ are the prices of commodities Q₁ and Q₂ respectively. Thus unreality is introduced by arbitrarily setting the quantities of commodities Q₃ through Q_n at zero. Entrepreneurial alertness which might be exercised through discovering additional arguments is assumed away and ignored. The price vector facing the consumer is assumed to be given. Thus any entrepreneurial activity aiming at improving the price vector from the consumer's perspective is also ignored. The consumer seeks to maximize the subjective, unobservable, ordinal utility function

U = f(q₁, q₂)

where q₁ and q₂ are the quantities of Q₁ and Q₂ the consumer consumes, subject to the budget constraint (Debreu 1959: 55-58, Silberberg 1979: 214-233, Henderson and Quandt 1980: 8-18, Chiang 1984: 400, Varian 1984: 113-115, Takayama 1985: 179-183.) For generality, no explicit functional form (such as U_t = Ac_t^a l_t^b) is assumed. (Varian 1984: 128-130 discusses explicit functional forms, including Cobb-Douglas and constant-elasticity-of-substitution utility.)

Because the utility function is assumed to be fixed, no scope is allowed for the entrepreneurial consumer to extract additional well-being from equal amounts of the same consumption goods. However, imagine a household budgeting for a fixed amount of fixed-price consumption goods over a period of time. A household which experimented with new food recipes could enjoy higher well-being by discovering recipes they prefer, (or recipes they dislike and avoiding them). This would shift the utility function closer to the origin. The optimization paradigm ignores this possibility of entrepreneurship in consumption. Entrepreneurial consumers would also seek to negotiate lower prices for consumer goods, and higher household incomes (Rothbard 1962: 183-200). This entrepreneurial behavior is all ignored by convention. Conventionally, the Lagrangean is

V = f(q₁, q₂) + λ(y⁰ - p₁q₁ - p₂q₂)

where λ is the as yet undetermined Lagrange multiplier. First-order conditions for an optimum are obtained by setting the first partial derivatives with respect to q₁, q₂, and λ equal to zero:

∂ V	=	f1 – λp1 = 0
∂ q1
∂ V	=	f2 – λp2 = 0
∂ q2
∂ V	=	y^0 - p1q1 - p2q2 = 0
∂ λ

Transposing the second terms in the first two equations and dividing yields the condition that the marginal utilities must be equal to the price ratio for a maximum<