Simple models of a human-capital based firm: a reference point approach

Paul Walker
Abstract One important feature of the knowledge economy is the increased importance placed on human-capital, especially when dealing with the rm. We apply the reference point approach to contracts to the modelling of a human-capital based rm. First a model of rm scope is oered which argues that the organisation of a human-capital based rm depends on the types of human capital involved. Having a rm based on a homogeneous group of human capital leads to a dierent organisational form than that of a rm which involves a heterogeneous group of human capital. Second a simple model of a human-capital based rm is discussed. Three organisational forms are considered: an investor-owned rm, a labour-owned rm and a market transaction involving the use of an independent contractor. Results are given that show when each of these forms are optimal. The eects of a rms size and scope on organisation are considered as is the question of Why are there conversions from worker to investor ownership?

JEL Classication: D23, D86, L22. Keywords: knowledge rm, contracts, theory of the rm, reference points, human capital.
Thanks to Ronda Kantor, Alan Woodeld and an anonymous referee for comments on previous drafts of this paper. psw1937@gmail.com.

introduction

The importance of knowledge to production and rm organisation has long been recognised. In 1888 Edwin Cannan noted, with regard to knowledge and production, that, The rst, and perhaps the most important, of the three causes which have led to the increase of the productiveness of industry is increase of knowledge. As regards this cause it is scarcely necessary to say anything. Everyone can see how enormously the productiveness of industry has been increased by the growth of mens knowledge of mechanics, chemistry, electricity, and other departments of science. In consequence of the growth of knowledge a few men can now do not only what it used to require many men to do, but also what could not formerly have been done by any number of men in any length of time (Cannan 1903: 12-3). while Alfred Marshall wrote two years later, with regard to knowledge and organisation, that Capital consists in a great part of knowledge and organization: and of this some part is private property and other part is not. Knowledge is our most powerful engine of production; it enables us to subdue Nature and force her to satisfy our wants. Organization aids knowledge; it has many forms, e.g. that of a single business, that of various businesses in the same trade, that of various trades relatively to one another, and that of the State providing security for all and help for many. The distinction between public and private property in knowledge and organization is of great and growing importance: in some respects of more importance than that between public and private property in material things; and partly for that reason it seems best sometimes to reckon Organization apart as a distinct agent of production (Marshall (1920) Book IV Chapter 1 page 115). Yet despite this early recognition the relationship between knowledge and organisation has received little attention in the contemporary theory of the rm. A range of issues related to the knowledge rm and entrepreneurship in the knowledge economy have been examined empirically and operationally - see for example Klein, de Haan and Goldberg (2010), Lin, Yeh and Hung (2012), Vliamos and Tzeremes (2012), Petrakis and Kostis (forthcoming) and Al-Omari, Al-Shaki, Ahmad and Ahmed (forthcoming) - but consideration of the theory of the knowledge rm has been much less.

The aim of this paper is to provide a theoretical foundation to enable the analysis of some of the claims that have been made about the eects that the increasing importance of knowledge, and the knowledge worker, to the production process will have on the organisation of the rm. To illustrate how having a rm based on knowledge workers - human capital - rather than physical capital can change the structure of the rm consider the example of pirates (a human-capital based rm) versus privateers (a more physical-capital based rm). Lesson (2009) argues that given the economic environment and incentives faced by pirates a worker cooperative type organisational form was found to be viable for pirate rms (ships), but as he also notes one size does not t all and thus worker-owned rms are unlikely to be an exploitable organisational form for all rms, in all situations.1 Leesons point about dierent economic contexts resulting in dierent organisational forms for rms can be illustrated by comparing the organisational form of privateers with that of pirates and considering the role that non-human capital (or the lack of it) plays in determining that form. An important dierence between privateers and pirates is that although they both practised maritime plunder, privateers were state-sanctioned. That is, governments would commission privateers to attack and seize enemy nations merchant shipping during times of war. The most obvious piece of non-human (physical in this case) capital for both pirates and privateers was their ship. The role of investors in providing this capital was important to the organisational form that pirates and privateers developed. Pirates had no investors; they simply stole the capital they needed. Privateers, on the other hand, as legal enterprises could not just go out and steal the capital they required, they needed external nanciers to supply their capital requirements. This dierence in capital supply resulted in very dierent organisational forms, the privateers having a more autocratic management system than pirates. Pirate crews were equal contributors and part owners of the rm they worked for.2 Having no need for investors, pirates did not need to develop mechanisms to protect the interests of the rms nanciers as the privateers needed to do. This meant that incentive problems could be dealt with by developing a worker-owned rm with the crew (usually) sharing equally in prot and electing its leaders and having power dispersed among multiple members of the crew such as the captain and the quartermaster. In contrast the privateer had investors and a management system designed to protect their investments.
1 But the sensibility of pirates democratic managerial organization in the particular context in which they operated doesnt mean democratic management makes sense for all rms in all circumstances. Dierent rms that operate in dierent economic contexts will nd dierent managerial forms most conducive to making prots (Leeson 2009: 182). 2 Bylund (2010) argues that Leeson is wrong to see pirate ships as rms. Bylund argues that the modern theories of the rm are not fully compatible with the nature of piratical organisation. Leeson replies in Leeson (2010).

The investors appointed privateer captains and developed an organisational scheme that in some important respects mirrored the managerial organisation of (also investor backed) merchant ships. If, as the above argument would suggest, the amount and nancing of the non-human capital required by the rm is an important determinant of a rms organisational form, Are rms going to resemble pirates, at least as far as they will be worker-owned, as human-capital becomes the dominate input to production?3 Or put it more generally, What determines the organisational form of the human-capital based rm? In the following sections we utilise simple versions of models developed from the reference point theory of Hart and Moore (2008) to study these questions.4 The models considered below dier from the standard theories of the rm in that the emphasis here is on human capital, important in the knowledge economy, rather than non-human capital which is central to the established approaches to the rm, developed with a more non-human capital based economy in mind. An obvious example of the non-human capital emphasis of the orthodox theories is the Grossman-Hart-Moore (GHM) approach in which the rm is dened as a collection of assets over which certain agents have property rights. (Moore 1992: 494). But this concentration on non-human capital means that none of the standard theories deal satisfactorily with human capital. Walker (2010) argues that [r]elationship-specic investmentinduced hold-up arguments for vertical integration are at their weakest when dealing with human capital. Human capital cannot be owned in the same way as physical capital and so the investor in human capital can act opportunistically whether an employee or not. The incentive-system theory assumes the use of a physical asset rather than a human capital asset in the production process. Neither the ownership nor the value of a human asset can be transferred and so such an asset cannot determine where the boundaries of the rm lie within the model. The extensions of the GHM framework oered by Brynjolfsson (1994) and Rabin (1993) inherits the implicit ownermanager restriction of the original GHM framework and thus are of limited use when modelling the knowledge rm. [ . . . ] The knowledge-based theory of the rm deals no better with the increasing importance of knowledge in the economy. (Walker 2010: 27). Thus the need for a new theory of the human-capital based rm for the knowledge economy. The rest of the paper is organised as follows. In Section 2 we examine how the type of human capital matters for the structure of the rm. We show that the organisation of the rm can depend on the nature of the human capital utilised by the rm. Having shown this Section
is an idea suggested by some commentators. See, for example, Rajan and Zingales (2003: 90). Walker (forthcoming) for a short survey of the application of the reference point approach to the theory of the rm.
4 See 3 This

3 puts forward a simple model of a human-capital based rm based around the reference point approach to the theory of the rm. Questions considered in this section include, Does the size and/or scope of the rm matter for the rms organisational form and Why are there conversions to investor ownership? Section 4 is the conclusion.

types of human capital and rm structure

Obviously the more a rm does in house, the greater the scope of the rm. Thus a rm which utilises employment contracts to keep production activities within its boundaries has a greater scope than a rm which hires independent contractors to carry out the same set of activities. In this section a model of the scope of a human-capital based rm will be considered. This section examines a simple model of a (predominately) human-capital based rm in which the owner of some non-human capital wishes to produce a widget that requires the input of two forms of human capital. Predominately human-capital based since as Hart (1995: 56-9) argues some non-human capital is an essential part of a theory of the rm. But such non-human capital can be either hard assets such as machines, inventories, and buildings, or softer assets such as patents, client lists, les, existing contracts, or the rms name or reputation. Assets in this softer category are likely to become more important as human capital becomes of greater signicance in production. To see the importance of non-human assets consider a situation were rm A takes over rm B which consists entirely of human capital. The question this raises is, What is to stop the workers of rm B from quitting? Without some non-human assets acting as glue to keep the workers in place, the workers could just announce one day that they are now a new rm. Without some non-human assets it is not obvious what keeps a rm together and what denes authority within a rm. Thus both non-human and human capital is required for a theory of the rm.5 The important point for the models below is that emphasis is on the human capital. In this section we will extend the analysis of Hart and Holmstr om (2010) to the case of a human-capital based enterprise. Unlike Hart and Holmstr om we look at a vertical, rather than horizontal, relationship and allow for a choice between an employment contract, with the owner of some non-human capital as the employer and two dierent knowledge workers as employees, or
5 Management scholar Peter F. Drucker makes the related point that in a knowledge society the standard factors of production still have a necessary role to play in production but that role is secondary to the role played by knowledge: [i]n this society, knowledge is the primary resource for individuals and for the economy overall. Land, labor, and capital the economists traditional factors of production do not disappear, but they become secondary. (Drucker 1992: 95).

the use of an independent contractor contract, where the non-human capital owner hires the two knowledge workers as consultants on the production process. Hart and Holmstr om consider the choice between two rms being separate entities or integrated. As with Hart and Holmstr om the knowledge workers will receive two forms of payos, a contractors fee or a wage and a private payo. The private payo can be thought of as measure of the job satisfaction of the knowledge worker. In contrast to Hart and Holmstr om the private payments utilised here can depend on the attitude of the knowledge workers towards being an employee or an independent contractor, that is the private payos can depend on the preferences or type of each of the knowledge workers. We will see that the continent nature of the private returns is important as the preferences or types of the human capital matters in determining the organisational structure of the rm. We will denote the non-human capital owner by C and the two knowledge workers (human capital) by A and B . We consider two basic organisational forms, an independent contractor contract and an employment contract. An independent contractor relationship is said to exist if C hires the two knowledge workers to act as consultants on the widget production process. In terms of a make-or-buy decision this can be thought of as a market contract - a buy decision. An employment contract is said to exist where the owner of the non-human capital makes the two knowledge workers employees of her rm. This can be seen as a make decision. In this model the distinction between the independent contractor and the employee is based on who has the right to make a coordination decision. Consider a situation where the knowledge workers, A and B , are an engineer and a computer scientist, working on dierent aspects of the widgets production, who could integrate their computer systems to make the transfer of information easier; or agreement could be needed on coordination of marketing campaigns; or to use the Hart and Holmstr om (2010) example, A and B could be deciding on whether or not to adopt a common standard or platform for their technology or product. Under an employment contract the employer, C , has the right to determine whether or not coordination of the activities of A and B takes place while if A and B are independent contractors they have the right to make the decision for their rm. Figure 6.1 represents the employment relationship while Figure 6.2 shows the independent contractor relationship.

Makes coordination decision

B Employment relationship (1 rm) Figure 6.1.

Makes coordination decision

Independent contractor relationship (3 rms) Figure 6.2.

Following Hart and Holmstr om (2010), the coordination decision - e.g. integrate the computer systems or not - is a Yes(Y) or No(N) decision made by whoever has the rights to the decision. If the decision rights are allocated to A and B then coordination occurs if and only if both select Y. If C has the decision rights the she can force A and B to carry out whatever decision she chooses. Each of A, B and C has an owner and a manager-employee, and these roles will coincide for each rm if A and B are independent contractors. If an employment contract is chosen then both A and B s rms will become divisions of C s rm and C will be the integrated rms owner and A and B will be manager-employees of their respective sections. Two forms of payo are generated by A and B : a contribution towards the monetary prot generated by the production of the widget and a private (nontransferable) benet in the form of a level of job satisfaction for the manager of the rm.6 It will be assumed that the owner of the widget can divert all the prots from its production to herself. The private benets, on the other hand, always stay with the managers. Importantly the managers private benets function can dier depending on whether or not the manager is an independent contractor or an employee. For the case where the manager is an independent contractor we denote the private payos as while if he is an employee the private payos are referred to as . The idea is that we allow for the possibility that A and B may evaluate the private benets from being self-employed versus being an employee dierently based on their personality types and/or personal abilities.7
benets are measured in money. for example, McCelland (1961) and Beugelsdijk and Noorderhaven (2005) for discussions of the link between personality type and employment preferences. Fuchs-Sch undeln (2009) looks at the preference for being self-employed and nds that the self-employed report higher job satisfaction than the employed. Hartog, van Praag and van der Sluis (2010) compares the eects of cognitive and non-cognitive abilities on the performance (measured by income) of entrepreneurs and employees. Their results show a markedly dierent returns for entrepreneurs and employees. Bandiera, Prat, Guiso, and Sadun (2011) show that managers risk aversion and talent are correlated with the incentives they are oered and, through these, with the characteristics of the rms that hire them. So managers with dierent characteristics are matched with rms with dierent characteristics. Puri and Robinson (2013) looks the attitudes of entrepreneurs and how they dier as a group from others in the economy. The paper shows that entrepreneurs are more optimistic, more willing to bear risk, and more motivated by the nonpecuniary enjoyment of work than wage-earners.
7 See, 6 Private

Y Y Firm A N A: B: A: B:

Firm B A: B: A: B:

N 0,0 0,0 0,0 0,0 Firm A N Y A: B: A: B:

Firm B A: B: A: B:

N 0,0 0,0 0,0 0,0

vA , A vB , B 0,0 0,0

vA , A vB , B 0,0 0,0 Employee payos Table 7.2.

Independent contractor payos Table 7.1.

Table 7.1 represents the payos from the situation where A and B are hired as independent contractors while Table 7.2 shows the payos if A and B are employees. In both tables vi , i = A, B represents the change in prots contributed by the two rms/divisions from a change in outcome from one involving at least one N decision to the Y, Y outcome and i , i = A, B and i , i = A, B represents the change in the private benets of the two managers, under each of the contract types. As in Hart and Holmstr om (2010) it is assumed that these payos are nonveriable and are perfectly certain. The payos have been scaled so that both the monetary prots and private benets under non-coordination are zero for both A and B . The time line is given in Figure 7.1. This time line shows that the parties meet and an organisational form (contract) is selected after which the coordination decision is made and later still payos are realised.

Organisational form chosen

Decision made

Payos realised

Figure 7.1. Time line. If an integrated rm is selected at the stage when the organisational form is determined then the two managers will work under a xed-wage employment contract and they will each receive the sum of the wage and the private benets of their section. If the use of independent contractors is the chosen organisational form then the two consultants will each receive a xed consulting fee in addition to the private benets generated by their rm. Under both organisational forms the prots from the production of the widget, vA + vB , goes to C since she is the owner of the widget under both contracts. It will also be assumed that as far as the change in prots is concerned

vA + vB > 0 7

(1)

holds.8 In terms of signs on the changes in private benets ve cases will be considered:9 A 0, A 0 Case 1 B 0, B 0 A 0, A 0 B 0, B 0 0 , 0 A A B 0, B 0 A 0, A 0 B 0, B 0 0 , 0 A A B 0, B 0 Case 2

Case 3

Case 4

Case 5

First, given that it has been assumed that inequality (1) holds we know that C wants coordination, that is, C wants, for example, the computer systems of the engineer and the computer scientist to be integrated. If Case 1 applies then the preferences of A, B and C are aligned. That is, they all want coordination under either the employment contract or the independent contractor contract. Thus no one is aggrieved and there is no shading.10 Therefore we get the rst-best meaning no aggrievement and no shading and therefore no deadweight loss outcomes under both the employment contract and the independent contractor contract. This means that social surplus is given by SE or SIC
8 This

= =

vA + vB + A + B vA + vB + A + B

(2) (3)

just says prots from production of the widget are positive. If they were negative production would not take place. 9 Consideration of the other logically possible cases does not alter the nature of the conclusions. 10 Very briey the Hart and Moore (2008) reference point theory argues that when the parties meet at date 0 there is uncertainty about the state of the world. This uncertainty is resolved shortly before date 1. There is symmetric information throughout, but the state is not veriable. A date 0 contract serves as a reference point for the contracting parties feelings about entitlements at date 1. Specically, neither party feels entitled to an outcome outside those permitted by the contract but within the contract there can be disagreement about the appropriate outcome. To simplify matters, it is supposed that each party feels entitled to their best possible outcome permitted by the contract. Of course, this means that usually at least one party will be disappointed or aggrieved by any particular outcome. Hart and Moore assume that no outcome from a transaction is perfectly contractible even at date 1. In particular, they assume that each party has the discretion to provide perfunctory performance rather than consummate performance. Performing at the lower perfunctory level rather than the higher consummate level is referred to as shading and it is assumed that shading cannot be penalised by a court. A court can, however, enforce the perfunctory level of performance. When a party is aggrieved he shades, by an amount times his level of aggrievement where 0 < 1. Consummate performance does not cost signicantly more than perfunctory performance, and a party will provide it if he feels well treated. Shading hurts the other party and causes a deadweight loss.

where SE is the social surplus under the employment contract and SIC refers to the social surplus under the independent contractor organisation. Which organisational form is chosen depends on which of equations (2) or (3) is the larger, which in turn depends on the relative sizes of A +B and A + B given that lump-sum transfers are available to distribute social surplus at the time the organisational form is chosen. Next assume that inequality (1) holds and Case 2 applies. Again C wants coordination, but both A and B do not, no matter which organisational form is chosen. Under an employment contract C can choose coordination and A and B must comply. That is, once C has made the Y/N decision the courts can and will enforce it. However, given Case 2 applies, both A and B will be aggrieved, and thus both will shade. Total shading will be (A + B ). C will therefore change her choice to no coordination if (A + B ) > vA + vB . In this case C reluctantly agrees to no coordination and is aggrieved and thus shades by an amount (vA + vB ). Thus an employment contract results in social surplus being given as

SE

vA + vB + A + B + (A + B ) if coordination occurs (vA + vB ) if coordination doesn t occur

(4) (5)

Note that the condition for coordination taking place is vA + vB + (A + B ) 0. Comparing this to the condition for rst best coordination, vA + vB + A + B 0, we see that there is too much coordination. That is, we can have coordination taking place when the payos from coordination dont justify it.11 Under an independent contractor contract both A and B will prevent coordination by choosing N . This will aggrieve C and she will therefore shade in total by an amount (vA + vB ). If we assume that C shades against A and B equally then A would change his decision to Y if
1 2 (vA 1 + vB ) > A and B will change his decision if 2 (vA + vB ) > B . If both

these inequalities hold then A and B will choose coordination, but only reluctantly which means they will be aggrieved and thus will shade.12 A shades by an amount A and B by B .
11 v + v + ( + ) > v + v + + > 0, since 0 < < 1 and 0 i = A, B , which i A B A B A B A B implies that it is possible to nd values such that vA + vB + (A + B ) > 0 vA + vB + A + B which says that rst best coordination will not take place but coordination will take place under the assumed conditions. 12 If only one of these inequalities hold then this is enough to ensure that coordination does not occur.

This determines that the social surplus will be given by

SIC

= vA + vB + A + B + (A + B ) if coordination occurs (vA + vB ) if coordination doesn t occur

(6) (7)

Under Case 2 neither the employment rst-best, equation (2), nor independent contractor rstbest, equation (3), can be achieved. The relevant comparison here is between equations (4)/(5) and equations (6)/(7). Here the conditions necessary for coordination to occur imply that (vA + vB ) + A + B > 0. Comparing this condition to the condition for rst best coordination, vA + vB + A + B > 0, we see there is too little coordination. That is, coordination may not take place even though the payos from coordination justify it occurring.13 If coordination does not occur, then SE = SIC while if coordination does occur which organisational form provides the greater surplus depends, again, on the relative sizes of A + B and A + B . Note that the employment contract errs on the side of too much coordination while the independent contractor contract errs on the side of too little coordination. Now assume that inequality (1) holds and Case 3 applies. In this case C wants coordination. To start with the employment contract, C can choose coordination and A and B must comply. However given that Case 3 applies, both A and B will be aggrieved, and thus they will both shade resulting in total shading of (A + B ). C will change her decision to N if (A + B ) > vA + vB . If this condition holds C will reluctantly agree to the no coordination outcome but is aggrieved and thus shades by an amount (vA + vB ). This gives the social surplus as

SE

vA + vB + A + B + (A + B ) if coordination occurs (vA + vB ) if coordination doesn t occur

(8) (9)

Note that the condition for coordination taking place is vA + vB + (A + B ) 0. Comparing this to the condition for rst best coordination, vA + vB + A + B 0, we see there is too much coordination, as in Case 2. In the independent contractor case the preferences of all of A, B and C are aligned, they all want coordination. There is no aggrievement and thus no shading. This means that social
13 Given that 0 < < 1 and v + v > 0, it must be that v + v + + > (v + v ) + A B A B A B A B A + B > 0. This shows that it is possible for vA + vB + A + B > 0 > (vA + vB )+ A + B which implies that rst best coordination would take place even if coordination under the assumed condition does not.

10

surplus is given by

SIC

= vA + vB + A + B

(10)

which is the independent contractors rst best outcome. In this case the independent contractor organisational form achieves optimal coordination in addition to the rst best outcome. Under Case 3 the employment rst best, equation (2), cannot be achieved, the best an employment contract can accomplish is (8)/(9). When Case 4 applies and inequality (1) holds the employment contract archives the rst best, as in Case 1: all preferences are aligned, they all want coordination. Social surplus is given by

SE

vA + vB + A + B

(11)

Under an independent contractor contract both A and B will prevent coordination, which C wants, as is the situation in Case 2. Case 2 also shows that social surplus is

SIC

= vA + vB + A + B + (A + B ) if coordination occurs (vA + vB ) if coordination doesn t occur

(12) (13)

We also see that under these conditions there is too little coordination, again as explained in Case 2. In this case the employment contract achieves the optimal level of coordination and results in the rst best outcome. For the last scenario it will be assumed that inequality (1) holds and Case 5 applies. Again C will favour coordination while A and B have diering views of the desirability of coordination: B wants coordination while A does not.14 In the case of the employment contract C will choose Y. This is in line with B s preferences and so he is not aggrieved and does not therefore shade. A, on the other hand, is aggrieved and shades by an amount (A ). This suggests that C will change her decision to N if (A ) > vA + vB . But it must be remembered that the choice of N will aggrieve B thereby leading to shading of (B ). So the condition for C to change her choice to N must take B s (potential) shading into account.15 The condition is therefore (A ) > vA + vB + (B ). If this condition holds C is aggrieved and shades (vA + vB ) and B is also aggrieved and shades by an amount (B ). This results in social
14 Sidepayments from B to A to induce A to accept coordination cannot be made due to the non-veriability of the private benets. 15 An assumption here is that if B shades he shades (fully) against C and not against A.

11

surplus being given as

SE

= vA + vB + A + B + (A ) if coordination occurs ((vA + vB ) + (B )) if coordination doesn t occur

(14) (15)

Under the employment contract the condition for coordination can be written

vA + vB + (A + B ) > 0

(16)

To determine whether there is too much or too little coordination this inequality must be compared to vA + vB + A + B > 0 (17)

The relationship between inequalities (16) and (17) is determined by the sign on the A + B term. That is, given that 0 < < 1
>

vA + vB + (A + B ) if and only if A + B

< vA + vB + A + B > 0
<

which means, depending on the sign of A + B we can get either too much or too little coordination. Under the use by C of independent contractors, A and B will make the coordination decision. B will want coordination while A will not and as it requires both A and B to select Y for coordination to occur, it will not occur. This results in both B and C being aggrieved, and thus induces shading, against A, by both of them. C shades by an amount (vA + vB ) and B by an amount (B ). The condition for A to choose coordination is therefore, (vA + vB ) + (B ) > A . This gives social surplus of

SIC

= vA + vB + A + B + (A ) if coordination occurs ((vA + vB ) + (B )) if coordination doesn t occur

(18) (19)

Under the independent contractors contract the condition for coordination can be written

A + (vA + vB ) + (B ) > 0

(20)

12

To determine whether there is too much or too little coordination this inequality must be compared to vA + vB + A + B > 0 The relationship between the values of inequalities (20) and (21) reduces to
> >

(21)

(21) < (20) if and only if (1 )(vA + vB + B ) < 0

(22)

which depends on the relative sizes of vA + vB and A , given that vA + vB > 0 and A < 0. Thus both too much and too little coordination are possible. As to which organisational form results in the greater social surplus, for SE > SIC , assuming coordination occurs, equation (14) needs to be greater than equation (18), that is

vA + vB + A + B + (A ) > A + B + (A ) >

vA + vB + A + B + (A ) A + B + (A )

which may or may not hold depending on the relative size of A (< 0) and B (> 0) and A (< 0) and B (> 0). If coordination does not occur then SE < SIC requires,
> >

((vA + vB ) + (B ))

<
<

((vA + vB ) + (B )) (vA + vB ) + (B ) (B )

(vA + vB ) + (B ) > (B ) >

<

which depends on the relative sizes of B and B . Thus either organisational form could be optimal depending on the preferences of A and B . The results derived in this section suggest that the organisational form a human-capital based rm will take depends, at least in part, on the preferences of the knowledge workers - or the types of human capital - involved with the rm. Having a homogeneous group of human capital involved in a rm may well lead to a dierent organisational form than that found in a rm which involves a heterogeneous group of human capital. This is an issue examined in more depth in the following sections.

13

a simple model of a human-capital based rm

In the inaugural Coase lecture given at the London School of Economics and Political Science in February 200716 Oliver Hart illustrated, via a simple example, how the reference point approach to the theory of contracts, introduced in Hart and Moore (2008), can be applied to the standard non-human capital based theory of the rm. The model developed here is a more formal and slightly more complicated extension of the basic structure of the Hart model applied to case of the human-capital based rm. Here we analyse the production of a homogeneous widget, as opposed to Harts musical evening, involving two agents, one of whom, a capital owner, supplies an amount of non-human capital along with a small amount of complementary human capital while the other, a knowledge worker, supplies the majority of the necessary human capital inputs. Hart assumes that one of his agents values the musical evening while the other faces an eort cost of providing the music. In the current model the widget generates revenue via its sale and both agents face eort costs. The similarity between the Hart model and the model below is the application of the reference point theory to the analysis of both situations. Hart compares an employment contract with the use of an independent contractor while the current models compares two dierent employment contracts, the capital owner as employer and the knowledge worker as employer, and the use of an independent contractor. The analysis will demonstrate that depending on the values of the price of the widget and the two eort costs each of the three contracts can be optimal. Date 0
Parties meet

Date 1
Widget produced

Figure 14.1. In the model developed in this section we deal with a long-term relationship between an owner of non-human capital and a knowledge worker. The time line for the relationship is given in Figure 14.1. The parties meet and contract (select an organisational form) at date 0 and can transact, in the future, at date 1. We assume that there is at date 0 a perfectly competitive market for non-human capital and knowledge workers but this competition is much reduced at date 1. At date 1 it is supposed, for the most part, that the owner of the non-human capital and the knowledge worker are bilateral monopolists. We have what Williamson (1985: 61-3) calls a fundamental transformation occurring between dates 0 and 1. Here we do not model why this transformation occurs.
16 The

published version of the lecture appeared as Hart (2008).

14

In this section three dierent organisational forms will be compared to determine when each is an optimal way of organising the production of a widget. A simple make-or-buy type decision will be studied in so far as a choice between the use of an independent contractor or an employment contract will be made. There are two forms that the employment contract can take. Either the knowledge worker can employ the (ex)non-human capital owner or the non-human capital owner can employ the knowledge worker. employment contract investor-owned rm

owner non-human capital

market contract independent contractor

knowledge worker

labour-owned rm employment contract Figure 15.1. Thus we can think of there being three types of contracts as illustrated in Figure 15.1. The rst involves the use of an independent contractor under which the owner of the non-human capital hires the knowledge worker as a consultant on the production process. This can be thought of as a market contract - a buy decision. The second is where the owner of the non-human capital employs the human capital (the knowledge worker), an arrangement which can be thought of as a very simple model of an investor or capital-owned rm.17 The third is where the human capital employs the (ex)owner of the non-human capital which can be thought of as a very simple version of a labour-owned rm.18,19 These employment contracts can be seen as make decisions.
17 With regard to the dierence between capital and labour-owned rms Dow (2003: 5) writes [a] capitalmanaged rm (KMF) is dened as an enterprise in which ultimate control is allocated by virtue of, and in proportion to, capital supply, while a labor-managed rm (LMF) assigns control by virtue of, and in proportion to, labor supply. 18 Labour-owned rms often employ non-owning workers. A rm of lawyers (a partnership), for example, may employ a secretary or receptionist and/or have employee lawyers, none of whom have ownership rights. 19 The modern theory of labour-owned rms is normally dated from Ward (1958). For a neoclassical theory of the rm approach to the labour-owned rm see Ireland and Law (1982). For a discussion of labour ownership utilising more recent approaches to the theory of the rm see Hansmann (1988: Section 5, 1996: chapters 5 and 6) and Dow (2003). Jensen and Meckling (1979) and Pejovich (1992) discuss some of the problems with labourowned rms. For a discussion of the (in)famous Yugoslavian experience with labour managed rms see George (1993: chapter 4). George (1993: 62) concludes that the unfortunately Yugoslavian experience cannot used to illustrate the economic failure of labour managed rms. On the extensive Spanish experience with LMFs see Morales Guti errez et al (2008).

15

More precisely, we consider a situation in which the production of a standard good - a widget, for which there are two possible versions - at sometime in the future, date 1, requires a combination of some non-human capital, N HC , and a small amount of eort by its owner, along with a much more substantive expenditure of eort by a knowledge worker, KW . The twist is that when the parties rst meet, at date 0, there is uncertainty about which widget will be produced. Details about the widget are too complicated to be veriable by the courts and thus the nature of widget to be produced cannot be specied in a date 0 contract. At date 1, however, the choice of the widget becomes clear. It is assumed that there is symmetric information throughout so that details such as the price for which the widget can be sold, p, the eort cost of the owner of the non-human capital, eN H and the eort cost of the knowledge worker, eKW , are observable to both the non-human capital owner and the knowledge worker. But these are not observable to third parties and thus are not veriable and so state-contingent contracts cannot be written. The wage of an employee or consultants fee of an independent contractor is xed in the date 0 contact to avoid aggrievement due to arguments over these payments, and thus shading. An independent contractor contract will be said to exist if the non-human capital hires the knowledge worker as a consultant when producing the widget. The consulting contract will specify that the knowledge worker will have the right to decide on which widget is to be made, but the non-human capital will retain ownership of the widget and thus the revenues from selling it. The owner of the non-human capital has the residual control rights in this case. The knowledge worker will be paid a xed consultants fee, w. An integrated rm will be said to exist under two conditions. The rst of these is when the non-human capital employs the knowledge worker to produce the widget. In this case the non-human capital will decide on the widget to be produced and keep the widget and its revenue while the knowledge worker will be paid a xed wage, w. Again the non-human capital owner has the residual control rights. The second form of an integrated rm is one where the knowledge worker employs the (ex)non-human capital owner and owns or rents the non-human capital. Here the knowledge worker will decide on the widget to be produced and keep the widget and its revenue while the non-human capital will be paid a xed wage, w. Here the knowledge worker has the residual control rights. Table 17.1 outlines the price of the two widgets along with the eort costs for the knowledge worker and the non-human capital and the total surpluses generated in the two cases. The dual role that the N HC plays here, both as the owner of non-human capital and supplier of eort is similar to that played by the informed party in models 2 and 3 of Rabin (1993). Rabin extends the property rights framework of Grossman and Hart (1986) to include human capital in the form 16

of information by assuming that an agent has knowledge about how to make a rms production process more productive which they are willing to sell. The problem is if the information is not revealed before the agent is paid, a (potential) buyer may have little reason to believe the agent is truly well-informed, but if the agent reveals the information up front, the buyer could simply use the information without payment. In Rabins two models the informed party has a dual role in that she can supply useful information and after that is revealed she can still contribute to productivity in some other way. Here we can think of N HC s non-human capital as being complementary to the non-human capital; for example he may be the only person who knows how to operate some necessary piece of machinery; or at least the knowledge worker cannot operate it.

Widget 1 Price Eort cost NH Eort cost KW Surplus p+ eN H + eH + p eN H eH +

Widget 2 p eN H eH p eN H eH

Table 17.1. Price/Costs of widget production. The organisational form that the rm will take will depend on a make-or-buy constraint and the relative size of changes in the eort costs of the owner of the non-human capital and the knowledge worker. Results 1 - 8 show the determination of the form of the rm for the case where > 0 and > 0.20 Result 1 If > 0, > 0, > , > with > and > + then ownership of the rm by the owner of the non-human capital is the optimal organisational form. These conditions can be interpreted as saying that switching from widget 2 to widget 1 will raise the eort costs for both the human and non-human capital as well as increasing the price of the widget. Also the increase in the eort cost for the non-human asset owner is greater than the increase for the knowledge worker. This means that should an employment contract be selected it is optimal to give ownership of the rm to N HC since her aggrievement level is greater than that of KW and therefore N HC cares more about the widget choice than KW . Here the ecient choice of widget is widget 1. > + is the make-or-buy condition for this
20 Obviously there are a number of other possible cases that could be looked at. But these follow the same logic as the cases dealt with here and thus can be worked through easily following the line of argument used here.

17

case. This condition determines whether an independent contractor or an employment contract will be used. Having a greater surplus for widget 1 means that ownership by N HC , who selects (the ecient) widget 1, dominates the independent contractor organisational form, under which widget 2 is selected. Proof: See Appendix Result 2 If > 0, > 0, > , > with either > and <
+ 1+

+ or < and

< + then an independent contractor contract is the optimal organisational form. Result 2 gives conditions under which integration does not take place. If this result holds then having two separate rms is optimal. From the proof of result 2 we know that either of the two stated two conditions is sucient for widget 2 to be the optimal choice and it is the independent contractor who selects widget 2. The rst condition can be seen as saying that an the independent contractor is preferred to a labour-owned rm while the second states that the independent contractor is preferred to an investor-owned rm. This makes sense since in the rst case > tells us that if an employment contract was to be used having the human capital as the employer would be optimal while in the second case if an employment contract was utilised having the non-human capital as the employer would be optimal as > . The intuition here is that both the make-or-buy constraints imply that widget 2 is the ecient widget to produce but as > and > both the labour-owned rm and the capital-owned rm will produce widget 1. However as the independent contractors eort costs are lower for widget 2 the ecient widget would be produced under this form of contract. Proof: See Appendix Result 3 If > 0, > 0, > , > with > and >
+ 1+

+ then ownership of the

rm by the owner of the human capital is the optimal organisational form. The condition > tells us that if an employment contract is used having the human capital as employer is optimal since the knowledge worker has higher levels of aggrievement and shading and thus generates greater deadweight losses. To avoid these losses ownership should be given to the knowledge worker since he cares most about the selection of the widget. The >
+ 1+

condition is the make-or-buy constraint which says that an employment contract is preferred to the use of an independent contractor. Proof: See Appendix

18

Result 4 If > 0, > 0, < , > then ownership of the rm by the non-human capital is never optimal. We have that < < , which says that if an employment contract is optimal then the knowledge worker should be the employer. The alternative to an employment contract is an independent contractor. Either way ownership by the non-human capital is not optimal. Proof: See Appendix Result 5 If > 0, > 0, < , > then use of an independent contractor is never optimal. < < + which implies the use of an independent contractor is optimal but only if the alternative employment contract has the non-human capital as the employer. But with > having the human capital as owner would be optimal. Proof: See Appendix Result 6 If > 0, > 0, < , > then human capital ownership is optimal. This is implied by Results 4 and 5. Proof: See Appendix Results 3 - 6 reinforce the idea that ownership by the knowledge worker makes sense when the relative important of the human capital is high. The fact that > indicates that in a switch from widget 2 to widget 1 the eort costs of the knowledge worker increases by more that those of the non-human capital and therefore the aggrievement of the knowledge worker is greater than that of the non-human capital and thus the shading potential of the knowledge worker is also greater than that of the non-human capital. Ownership is therefore placed in the hands of the group which has the greatest interest in the choice of the widget. This suggests that if the rm has a homogeneous group of knowledge workers whose interests are highly correlated, and thus their reasons for aggrievement and levels of aggrievement, in the model, are similar, then labour ownership is likely to be a stable organisation form. Dow (2003: 251) makes the point that the congregation of labour-owned rms in areas in which there is limited division of labour and the skill sets of workers are similar is due, in part, to having to overcome problems related to heterogeneity of preferences: [t]he heterogeneity of worker preferences can account for the tendency of LMFs [labour-managed rms] to congregate in industries where workers have similar skills and rms have a limited division of labor. This is a feature of most traditional workers 19

cooperatives and many professional partnerships. [ . . . ] Asymmetries in the internal composition of control groups can thus explain various observations about the design, incidence, and life cycles of LMFs that would be dicult to rationalize in terms of nancial constraints alone. The restriction to workers with similar skills with little utilisation of the division of labour reduces the heterogeneity of worker preferences, e.g. the workers are all of type , and this reduces asymmetries in the composition of the controlling group thereby reducing aggrievement and shading.21 Hansmann (1996: 91) asks the question, with regard to employee-owned rms, Which Firms Succeed? In part his answer is, [t]he most striking evidence of the high costs of collective decision making [for employee-owned rms] is the scarcity of employee-owned rms in which there are substantial dierences among employees who participate in ownership. Most typically, employee-owned rms all do extremely similar work and are of essentially equivalent status within the rm. Rarely do they have substantially dierent types or levels of skills, and rarely is there much hierarchical authority among them. When considering what constitutes homogeneity of interest for employee ownership Hansmann writes, [t]he preceding evidence implies that employee ownership works best where the employee-owners are so homogeneous that any decision made by the rm will aect them roughly equally, or where, though the employees dier in ways that cause the burdens and benets of some decisions to be shared unequally, there is an objective and widely accepted basis for making those decisions. That is, employee ownership is most viable where either no important conicts of interest exist among the employeeowners, or some simple and uncontroversial means is available to resolve the conicts that are present. (Hansmann 1996: 97). Such observations are consistent with the basic results of this section. What we gain from the results above is an understanding, in terms of reducing aggrievement and shading, of why homogeneity is important.
21 Having team members of a similar types also makes mutual monitoring less costly as each worker knows what the others should be doing and thus reduces opportunism.

20

If we increase the heterogeneity of the rms workforce by increasing the relative importance of the non-human capital owner in the model, then the optimal organisational form will move away from a labour-owned rm. See Results 1 and 7. Result 7 If > 0, > 0, > , < then independent contractor and non-human ownership result in the same ecient outcome and both dominate human capital ownership. Proof: See Appendix Result 8 If > 0, > 0, < , < then independent contractor, non-human ownership and human ownership all result in the same ecient outcome. Result 8 gives us the conditions under which all three organisational forms select the ecient widget, widget 2, and there is no shading in any case and thus all three result in the same total surplus. With no deadweight loss generated, ownership ceases to matter.22 Proof: See Appendix Thus we see that the reason that ownership of the rm by the non-human capital is possible is due to the presence and size of the eort costs of the owner of the non-human capital. Without this there would be no shading under human capital ownership and thus control of the rm by the knowledge worker would be ecient and therefore the optimal choice of organisational form. This can be interpreted as suggesting that if human-capital based rms are made up of a homogeneous group of knowledge workers then some form of worker ownership is optimal. But if the rm introduces more heterogeneity into its workforce - in the simple model used here, the inclusion of eort by the capital owner in the production process - then ownership by the nonhuman capital becomes possible. As already noted this is consistent with rms like partnerships normally consisting of a single group of knowledge worker, e.g. lawyers, doctors but not both.

3.1

does size matter?

Brynjolfsson (1994) argues that small rms have advantages in the provision of incentives for the exploitation of information. In Brynjolfssons view small rms therefore have an advantage over larger ones in situations in which it is important to provide incentives for the application of information in ways that cannot be easily foreseen and incorporated into a contract. Brynjolfsson (1994: footnote 12) goes further by noting that the stronger, output-based incentives
22

< implies that both inequalities 23 and 24 hold.

21

for the non-contractible actions in smaller rms will not only induce higher eort overall, but in multidimensional models, will also induce less eort on actions that do not enhance output. In the model of the current section the role of size is to help determine the ownership structure of the rm. Insofar as smaller rms are more homogeneous, some form of labour ownership is more likely. Partnerships, for example, tend to be small. If as a rm grows so does the heterogeneity of that rm then the possibility of aggrievement and thus shading resulting in deadweight losses also grows. If this is interpreted as an increase in for the current model, then the likelihood of non-labour ownership increases. This also implies that investor-owned human-capital based rms will tend to be larger than similar rms that are labour-owned.

3.2

does scope matter?

Kling (2010: 47) makes the argument that as knowledge becomes more specialised the scope of rms will tend to decrease, that is, the range of activities carried out by an individual (knowledgebased) rm will decrease. Being highly competent in one specialised area means that it likely that you will have to make decisions outside your area of expertise if the rm has broad scope. If, as seems reasonable, a decrease in activities is associated with a reduction in the heterogeneity of the types of knowledge workers employed in the rm then some form of labour ownership is more likely. Partnerships such as lawyers or accountants are the obvious example of where a form of labour ownership and limited scope coincide.

3.3

why are there conversions to investor ownership?

In raising this question Hansmann (1996: 83) writes [i]f [. . . ] there is no perverse mechanism that causes successful employee rms to convert to investor ownership simply as a consequence of their very success, then why do conversions from employee ownership to investor ownership occur so frequently? The answer suggested here would be that conversion can in fact be a result of the success of the rm, at least for human-capital based employee-owned rms. If success means growth of the rm and thus increased heterogeneity of the rm then conversion makes sense. In the model of this section an increase in heterogeneity can be interpreted as the introduction of, or increase in, which changes the relative sizes of and and thus could result in a change in ownership.23
23 Hansmanns answer is that [t]he most likely explanation is simply that employee ownership is not an ecient mode of organization for the rms involved. (Hansmann 1996: 83). Dow (2003: 225) oers two other reasons: employees may not longer need the insurance features of a LMF and for collective choice reasons investors may be able to design a takeover oer 51% of a LMFs members will accept.

22

A short summary of the results this section is that the organisational form a knowledge rm takes depends on two conditions: 1) a make-or-buy constraint and 2) the level of aggrievement that each of the possible owners feels.

conclusion

The shift toward a knowledge economy has placed an increased emphasis on the role of human capital in the production process and in the rms formed to facilitate that process. A question that this gives rise to is What organisational form will rms based on human-capital take? The models analysed in this paper suggest that the ownership structure that will develop depends on more than just the fact that the knowledge worker is the rms primary source of value added. It also depends on the nature and homogeneity of the human-capital involved in the rm. A frequently observed organisational form that human-capital based rms currently take is some form of labour-owned rm, a partnership being the most common example. As human capital has increased in importance some commentators have argued that knowledge based rms are increasingly developing organisational structures along the lines of partnerships. Rajan and Zingales (2003: 90) argue that, [h]uman capital is replacing inanimate assets as the most important source of corporate capabilities and value. In both their organizational structure and their promotion and compensation policies, large rms are becoming more like professional partnerships. One conclusion that follows from the analysis of this paper is that while there is truth to the Rajan and Zingales claim, it is not likely that their argument is universally applicable. While some rms may gravitate towards a more partnership like structure, others will maintain a more traditional organisational framework and still others may rely on outsourcing. The models of this article show that dierences in preferences or in the types of human capital matter for the choice of organisational form. From the rst model we see that when preferences are homogeneous the rst best can be achieved with the optimal level of coordination occurring and the scope of the rm being clearly dened. Heterogeneity of preferences results in both the level of coordination and the scope of the rms being dependent on the parameter values. We also found that deadweight losses can occur under these conditions.. In the second model it is shown that both outsourcing via the use of an independent contractor or in-house production via an employment contract are possible. With regard to the use of an 23

employment contract the more heterogeneous the human capital used by the rm the more likely is an investor-owned rm to be formed. The labour-owned rm is more likely to be formed when the human capital is relatively homogeneous in its characteristics and faces a common set of incentives. The pirate example at the beginning of the paper is an (unusual) example of this. Another (more usual) example of a human-capital based rm, but one where labour-owned rms are seldom, if ever, found, is that of the professional sports team. The models developed above can explain this. Here we have a situation where human capital, talent at playing a particular sport, is the basis for the rm but ownership by the human capital, the players, is extremely rare since a worker-owned team would be at a disadvantage relative to a player-asemployee based team. Heterogeneity in playing talent - playing talent being the human capital here - and thus in earning potential24 is a disincentive to the formation of a worker cooperative, an organisation which normally involves (rough) equality in payment,25 since those players with the greatest earning potential, the largest outside options, will transfer away from the cooperative to maximise their income stream.26,27,28 Thus a worker-owned team would have few, if any, star players, a handicap in the winner-takes-all world of professional sports.29
24 As Goddard and Sloane (2005: 353) explain [r]eal stars are few in number and possess skills that cannot be easily replicated by training the average player more intensively. Rosen (1981) referred to this as the superstar phenomenon, and illustrated how human capital interacted with production technology to magnify small dierences in talent, resulting in large dierences in earnings. [ . . . ] Characteristically, wages are highly convex with respect to star quality, resulting in a highly skewed distribution of earning with a long upper-tail to the distribution. 25 Some worker cooperatives do require all sta to be paid the same wages while others have a limit on the range of pay oered. Such limit is normally expressed as as ratio of the highest paid worker to the lowest paid, e.g. 3:1 or 4.5:1. Also an owners allocation of the prots of the rm will be proportionate to that owners labour input, so for similar labour inputs, owners receive a similar allocation of prots. Thus overall compensation is commonly roughly equal. 26 Ricketts (1999: 20) outlines the general problem as [. . . ] to minimise antagonism a rough equality in the division of the residual will be necessary and this may conict with outside opportunities. Those with high transfer earnings reecting high productivity elsewhere will desert the co-operative. It is for these reasons that control of the rm by its labour force is usually found in circumstances which permit a high degree of common interest. Jossa (2009: 709-10) explains the basic issue in terms of the management of capitalistic versus co-operative rms: [g]iven the tendency of cooperatives to distribute their income equitably among all the members, it is dicult to deny that few cooperatives are in a position to pay the high salaries that able managers can expect to earn in capitalistic rms. Whenever a group of people resolve to work as a team-we may add-the member who outperforms the others in initiative and organizational skills will inevitably take the lead. The crux of the matter is that such a person has no incentive to establish a cooperative and share power and earnings with others. He or she will prefer to found a capitalistic rm, where he or she will hold all authority and, if sole owner, appropriate the whole of the surplus [references deleted]. 27 One way around the problem is to pay all players the amount necessary to retain the best player. But this implies paying most players more than their productivities and has obvious implications for the teams budget constraint. This would also aggravate the problem of non-performing average player-owners not wanting to leave the team. 28 Dierential payment schemes can occur, especially within partnerships, but they require that the individual employee productivities are suciently easy to measure so that a relatively objective method of productivity related pay is possible. Given the team production nature of team sports productivities are dicult, if not impossible, to estimate and thus payment by productivity is not feasible, which argues in favour of equality in payments. 29 Scully (2008) notes that there is a relationship between a teams nances and a teams success since better nanced teams can buy the best players: [h]ow good a professional sports team is depends, of course, on the quality of its players. Because teams compete for better players by oering higher salaries, the quality of a team depends largely on how strong it is nancially. The nancially stronger teams will, on average, be the better teams.

24

Another issue for a cooperative sports team is that while the star players may leave the team too soon, the average players may stay too long. The average players will have smaller outside options and thus less incentive to leave but as they are also owners of the team it would be more dicult to get rid of those who are not performing. It would be easier for an employee-based team to remove under performing players as they are not owners of the rm. Also to the degree that exit barriers are entry barriers a worker-owned organisation is at a disadvantage. Such an organisation could hinder rapid transfers between clubs. Problems with transfers could arise, for example, if the condition under which a player can exit the team have to be negotiated with the remaining player-owners at the time of exit. Or the remaining owners may be unable or unwilling to buy out the exiting player under a right of rst refusal or right of rst oer scheme or any of them could veto an incoming replacement player-owner. Also there is the question of the value of a players interest in the team as well as the question of the time period over which an agreed upon value would be paid. These costs make exit more dicult than it would be under an employment contract and thus tend to lock-in the player-owner to the team. Such lock-in is a disincentive to forming, or joining, a labour-owned rm, especially for the best players.30 Many of these problems can, to a degree, be contracted around but this imposes additional negotiation costs at the time of entry into the team, which again is a disincentive to forming an worker-owned team. Utilising a worker-owned organisation results in additional haggling costs, either ex ante or ex post, relative to a player-as-employee team. Put simply, an employee can leave a organisation more quickly and easily than an owner and in the case of professional sports, transfers between teams, or at least a credible threat to transfer, are particularly valuable to the best players. Therefore a player-owned team would be at a competitive disadvantage compared to teams comprised of employee players. In the model of Section 3 an independent contractor contract implies giving control over production decisions (choice of the widget in the model) to the consultant which in the sports team example would mean giving control to the players. This would create at least some of the same problems as player ownership. If the consultants (players) contract restricted the amount of control that players have then its not clear what the eective dierence between the independent contractors contract and an employee contract would be.
30 There are also problems such as those which could arise between the manager or coach and the player/owners in their roles as players and as owners. In addition to this there is the problem that if the players provide any nancial capital the team needs themselves they risk being being badly underdiversied. If the team goes bankrupt they lose not only their job but also (at least part of) their savings. Given the short time span that a player is likely to be a member of the team there is the issue of the return on any capital the player has invested in the team. To ensure a suitable post playing standard of living players are likely to place an emphasis on high returns on their investments.

25

An alternative interpretation of the models developed above is that they show that organisational form depends, in part, on the relative mix of the inputs used in production. For a largely human-capital based rm some form of labour-ownership is feasible while for a rm with a greater (relative) use of non-human capital, a capital owned organisation is more likely. To see this note that within this framework an increase in , relative to , increases the relative importance of the non-human capital compared to the human capital. As the amount of non-human capital the knowledge worker needs to be productive increases, the ability of the non-human capital owner to shade, and thus his ability of impose welfare losses, also increases. Thus what we see is that when the rm is homogeneous in its inputs, in the sense that they consist largely of the eorts of the knowledge worker meaning is relatively small some form of worker-owned organisation is feasible. But as the relative importance of the non-human capital increases, that is, the heterogeneity of inputs increases with becoming relatively large, the rms begins to look more traditional in its input mix and thus begins to look more traditional in its organisational form in that a capital-owned rm becomes increasingly likely. If you are a computer scientist using a PC to design small business accounting software then your rm being a partnership is feasible whereas if you are writing software for a super-computer you are more likely to be an employee of whoever owns/rents the computer. A somewhat more unorthodox interpretation of the Section 3 model arises if we are to think about the model as one without non-human capital, that is, a model with just two types of human capital and thus both and would represent eort by knowledge workers. First, consider what happens if and are approximately the same. Here it is not clear who the owner should be as each type can shade to, roughly, the same degree, albeit for dierent reasons. This could make the ownership of the rm indeterminate and the rm seemingly unstable. What is to stop one or other of the knowledge workers from just announcing one morning that they have become a new rm? Such a result would be consistent with the Hart (1995: 56-7) argument that even a human-capital based rm needs some non-human capital to act as glue to keep the rm together. Next assume that one of either or is larger than the other. Here the results of Section 3 could be interpreted as showing why human-capital only rms tend to be owned by a single type of human capital. Ownership would be allocated to whichever of the types of human capital is most likely to be aggrieved and thus most able to shade. This makes sense as it minimises any possible deadweight losses. The problem with this is the same as in the previous case, the concept of the rm is not well dened. Again it is not clear why the rm would stay together, 26

What is the glue in this case? What stops the owner from forming a new rm without the non-owner or the non-owner forming his own rm? What these last two paragraphs highlight is the importance of the non-human capital in the model. Without this it seems a rm is little more than a single knowledge worker; a phantom without any real organisation or structure. Thus we observe the most common form of worker ownership, partnerships, mostly in situations where the rms are small and the human input is of a single type, e.g. just lawyers or just doctors. Also the partners in the rm tend to stay with the rm for a long period which minimises the exiting problem. What this suggests is that a human-capital only rm with heterogeneous human capital is likely to be unstable and thus a long lasting human-capital only rm will consist of homogeneous human capital. A rm which involves heterogeneous human capital will require some glue, in the form of non-human capital of some kind, to remain viable. Given the importance of this glue to the rm ownership of the rm by the owner of the non-human capital is likely. In the models considered above size helps to determine ownership structure in that smaller rms are more likely to be homogeneous in their labour inputs and thus some form of labour ownership is more likely. If as rms grow so does the heterogeneity of their workforce then the possibility of aggrievement and thus shading resulting in deadweight losses also grows. The results above would suggest therefore that the likelihood of non-labour ownership increases. The scope of the rm has similar results for its organisational form. Being highly competent in one specialised area means that it likely that you will have to make decisions outside your area of expertise if the rm has broad scope. If a decrease in activities is associated with a reduction in the heterogeneity of the types of knowledge workers employed in the rm then some form of labour ownership is more likely. The papers arguments also suggest that both vertical and horizontal integration can promote economic eciency. Such eciencies need to be considered when assessing the likely eects of a merger. This paper aimed to provide a simple theoretical framework for the analysis of the eects of the increasing importance of human capital to the organisational structure of rms. Perhaps the most general insight oered is that it is not only the greater importance of human capital to rms that determines the form that businesses take but it is also the heterogeneity of that capital that helps determine a rms structure.31
31 This does raise the question of what homogeneity of interest is. Hannsmann (1996: 97-8) considers this question and notes that there is no unique answer. It follows that homogeneity of interest is, in important degree, a social construct. Consequently, what passes for homogeneity in one setting may not in another. (Hansmann

27

appendix: proofs of results

Proof of Result 1: First we compare the performance of the non-human capital owned rm, (N HC ), with that of the independent contractor. In the independent contractor case the knowledge worker, (KW ), acts as a consultant to N HC on the production of the widget. KW receives a xed fee, w. Given this xed fee it is optimal for him to pick widget 2 because of the lower eort cost associated with widget 2: eKW < eKW + as > 0. The payo for KW will therefore be w eKW . The payo of N HC is p eN H w. N HC will shade because his payo is less than that he would have received had widget 1 been chosen: p eN H w + > p eN H w given > . N HC shades by an amount ( ). The total surplus will therefore be p eN H w + w eKW ( ) = p eN H eKW ( ). In a non-human capital owned rm, N HC employs the knowledge worker to produce the widget. If N HC owns the rm then he will pick widget 1 as the payo is higher: p + eN H w > p eN H w since > . Here w is the (xed) wage recieved by KW . The payo for KW is therefore w eKW . KW will shade by amount since his eort costs are eKW + for widget 1 against eKW for widget 2 and > 0. The total surplus will therefore be p + eN H w + w eKW = p + eN H eKW . Thus the condition for the total surplus of a non-human capital owned rm to be greater than the total surplus of an independent contractor is

p + eN H eKW ( ) + ( ) ( )(1 + )

> p eN H eKW ( ) > ( ) > + > + (1 + ) 1+

>

> + .

In a human capital owned rm, KW employs the N HC to produce the widget. If KW owns the rm then he will pick widget 1 as the payo is higher: p + eH w > p eH w since > . The payo for N HC is w eN H . In this case w is the wage paid to N HC . N HC will shade by amount since his eort costs are eN H + for widget 1 against eN H for widget 2 and > 0. The total surplus will therefore be p + eH w + w eN H = p + eH eN H .
1996: 98).

28

Thus the condition for the total surplus of a non-human capital owned rm to be greater than the total surplus of an human capital owned rm is

p + eN H eKW

> p + eH eN H > < <

which is true by assumption. Thus for the given parameter values the surplus for the non-human capital owned rm is greater than either of the alternatives. Proof of Result 2: We know the equations for the surpluses of the three organisational forms from the proof of result 1. Thus we just need to show the conditions under which an independent contractor contract dominates the alternatives. For the independent contractor to dominate the non-human capital ownership we need:

p eN H eKW ( ) > ( ) > + >

p + eN H eKW ( ) + ( ) (1 + ) + . (23)

(1 + )( ) < <

Comparing the independent contractor with the human capital owned rm we get:

p eN H eKW ( ) > ( ) > + > (1 + )( ) < <

p + eKW eN H ( ) + ( ) + + + 1+ (24)

Depending on the relative values of and we can rank the righthand sides of inequalities (23)

29

and (24).
> + + < 1+

+ 1+

<
>

>

(1 + )

+ < < <

+ 1+ + 1+ > >

Thus if > then < + < means that either > and < (24).

+ and if < then <

+ 1+

+ < + . This

+ or < and < + will satisfy both (23) and

Proof of Result (3) : We know the equations for the surpluses of the three organisational forms from the proof of result 1. Thus we just need to show the conditions under which human ownership dominates the alternatives. First consider human ownership against the independent contractor.

p + eH eN H ( ) + ( ) (1 + )( )

> p eN H eKW ( ) > ( ) > + > + > + + . 1+

Next consider the case of the human capital owned rm against the non-human capital owned rm.

p + eH eN H > > <

p + eN H eKW .

Thus for the given parameter values the surplus for the human capital owned rm is greater than

30

the alternatives. Proof of Result 4: For the case of the independent contractor, KW will pick widget 2 as he receives a xed fee and costs are lower for him with widget 2: eKW < eKW + given > 0. KW s payo will therefore be w eKW . N HC s payo will be p eN H w. N HC will shade by an amount ( ) since his payo would be higher for widget 1: p eN H w + > p eN H w since > 0. Total surplus is therefore p eN H eKW ( ). If the non-human capital is the employer then he will choose widget 1 since his payo would be higher: p eN H + w > p eN H w given that > . N HC s payo is p eN H + w. KW s payo will be w eKW . KW will shade by an amount given his costs are higher for widget 1. Total welfare will be p eN H + eKW . N HC ownership will dominate IC if

p eN H + eKW (1 + )( )

> p eN H eKW ( ) > ( ) > (1 + )

> > +

But + given that by assumption it is less than . Under KW ownership widget 2 will be chosen since p eKW w > p eKW w + which implies < 0 which is true as > . KW s payo will be p eKW w while N HC s payo will be w eN H . N HC will not shade as his costs are less under widget 2 Total surplus will be p eKW w + w eN H = p eKW eN H . Thus HNC ownership will dominate KW ownership if

p eN H + eKW

> p eKW eN H > 0 > + + .

But this cannot be true since < by assumption.

31

Thus N HC ownership dominates neither of the other organisational forms. Proof of Result 5: IC cannot be optimal since it is dominated by KW ownership. For IC to dominate KW it must be that

p eN H eKW ( ) >

p eKW eN H

( ) >

which cannot be true since > . Proof of Result 6: For human capital ownership to be optimal it must be that it dominates both non-human ownership and the independent contractor. To show the rst part:

p eKW eN H 0

> p eN H + eKW > < + +

which is true because < alone. For the second part:

p eKW eN H 0

> p eN H eKW ( ) > ( )

which is true since > . Proof of Result 7: Given that > 0 for the case of the independent contractor, KW will pick widget 2 since he faces a xed fee and widget 2 has lower eort costs. KW s payo will be w eKW . N HC s payo will be p eN H w. There will be no shading by N HC since p eN H w > p eN H w + because > . Total surplus will be p eN H eKW . For the case where N HC is the employer he will pick widget 2 since p eN H w > p eN H w + given that > . KW will not shade since widget 2 is his least eort cost state. Thus the total surplus will be p eN H eKW . This shows that in this case the independent contractor case and the N HC employer case give the same ecient outcome. In the third case where KW is the employer, KW will pick widget 1 since p+ eKW w > p eKW w because > . KW s payo is p + eKW w while N HC s payo is w eN H 32

which is less than his payo for widget 2 which is w eN H because > 0. This means N HC will shade by an amount . Total surplus will therefore be p eN H eKW + . Given that < + + , since < by assumption, human capital ownership is dominated by both of the alternatives. Proof of Result 8: Given the xed consultants fee, KW will choose the least cost widget which is widget 2. being less that means N HC maximises his surplus with widget 2 and thus will not shade. As is less that both and there is no incentive for either employer to pick widget 1 and as widget 2 is least cost for both KW and N HC , neither will shade. As all three organisational types pick the ecient outcome and there is no shading under any organisation, all three organisational forms must result in the same level of total surplus.

References
Al-Omari, Zuher Hamed; Osama Treef Al-Shaki, Mohd Sharifuddin Ahmad and Elsadig Musa Ahmed (forthcoming). Knowledge Growth Measurement and Formulation for Enhancing Organizational Knowledge Capital, Journal of the Knowledge Economy. Bandiera, Oriana, Andrea Prat, Luigi Guiso and Raaella Sadun (2011). Matching Firms, Managers and Incentives, NBER Working Paper No. 16691, January. Beugelsdijk, Sjoerd and Niels Noorderhaven (2005). Personality Characteristics of Self-Employed; An Empirical Study, Small Business Economics, 24(2) March: 159-67. Brynjolfsson, Erik (1994). Information Assets, Technology, and Organization, Management Science, 40(12) December: 1645-62. Bylund, Per (2010). Piracy, Inc. on the bearing of the rm analogy to pirate organization, Review of Austrian Economics, 23(3) September: 299-305. Cannan, Edwin (1903). Elementary Political Economy, 3rd edn., London: Henry Frowde. First edition 1888. Dow, Gregory K. (2003). Governing the Firm: Workers Control in Theory and Practice, Cambridge: Cambridge University Press. Drucker, Peter F. (1992). The New Society of Organizations, Harvard Business Review, 70(5) September-October: 95-104. Fuchs-Sch undeln, Nicola (2009). On preferences for being self-employed, Journal of Economic Behavior and Organization, 71(2) August: 162-71. George, Donald A. R. (1993). Economic Democracy: The Political Economy of Self-Management and Participation, London: The Macmillan Press. Goddard, John and Peter J. Sloane (2005). Economics of sport. In Simon W. Bowmaker (ed.), Economics Uncut: A Complete Guide to Life, Death and Misadventure (chapter 12, pp. 345-66), Cheltenham U.K.: Edward Elgar Publishing Ltd.

33

Grossman, Sanford J. and Oliver D. Hart (1986). The Costs and Benets of Ownership: A Theory of Vertical and Lateral Integration, Journal of Political Economy, 94(4) August: 691-719. Hansmann, Henry (1988). Ownership of the Firm, Journal of Law, Economics, and Organization, 4(2) Autumn: 267-304. Hansmann, Henry (1996). The Ownership of Enterprise, Cambridge, Mass.: Harvard University Press. Hart, Oliver D. (1995). Firms, Contracts, and Financial Structure, Oxford: Oxford University Press. Hart, Oliver D. (2008). Economica Coase Lecture: Reference Points and the Theory of the Firm, Economica, 75(299) August: 404-11. Hart, Oliver D. and Bengt Holmstr om (2010). Contracts as Reference Points, Quarterly Journal of Economics, 125(2) May: 483-513. Hart, Oliver D. and John Moore (2008). Contracts as Reference Points, Quarterly Journal of Economics, 123(1) February: 1-48. Hartog, Joop, Mirjam van Praag and Justin van der Sluis (2010). If You Are So Smart, Why Arent You an Entrepreneur? Returns to Cognitive and Social Ability: Entrepreneurs Versus Employees, Journal of Economics and Management Strategy, 19(4) Winter: 947-89. Ireland, Norman J. and Peter J. Law (1982). The Economics of Labour-Managed Enterprises, London: Croom Helm. Jensen, Michael C. and William H. Meckling (1979). Rights and Production Functions: An Application to Labor-Managed Firms and Codetermination, Journal of Business, 52(4) October: 469-506. Jossa, Bruno (2009). Alchian and Demsetzs Critique of the Cooperative Firm Thirty-Seven Years After, Metroeconomica, 60(4) November: 686-714. Klein, Raphael, Uzi de Haan and Albert I. Goldberg (2010). Corporate Exploration Competence and the Entrepreneurial Enterprise, Journal of the Knowledge Economy, 1(2) June: 86-116. Kling, Arnold (2010). Unchecked and Unbalanced: How the Discrepancy Between Knowledge and Power Caused the Financial Crisis and Threatens Democracy, Lanham: Rowman & Littleeld Publishers, Inc. Leeson, Peter T. (2009). The Invisible Hook: The Hidden Economics of Pirates, Princeton: Princeton University Press. Leeson, Peter T. (2010). Pirates, Review of Austrian Economics, 23(3) September: 315-19. Lin, Chien Chiang; Ju Lie Yeh and Guan Wen Hung (2012). Internal Impediments of Organizational Innovation: An Exploratory Study, Journal of the Knowledge Economy, 3(2) June: 185-98. Marshall, Alfred (1920). Principles of Economics, 8th edn., London: Macmillan and Co. Ltd. First edition 1890. McCelland, D. (1961). The Achieving Society, Princeton: Van Nostrand Reinhold. Moore, John (1992). The Firm as a Collection of Assets, European Economic Review, 36(2-3) April: 493-507.

34

Morales Guti errez, Alfonso Carlos; Sonia Mart n L opez and Gustavo Lejarriaga P erez de las Vacas (2008). Labour Managed Firms in Spain, CIRIEC-Espa na, Revista de Econom a P ublica, Social y Cooperativa, N um. 62, octubre-sin mes: 61-85. Pejovich, Svetozar (1992). Why Has the Labor-Managed Firm Failed, Cato Journal, 12(2) Fall: 461-73. Puri, Manju and David T. Robinson (2013). The Economic Psychology of Entrepreneurship and Family Business, Journal of Economics & Management Strategy, 22(2) Summar: 423-44. Petrakis, P. E. and P. C. Kostis (forthcmoning). The Role of Knowledge and Trust in SMEs, Journal of the Knowledge Economy. Rabin, Matthew (1993). Information and the Control of Productive Assets, Journal of Law, Economics, and Organization, 9(1) Spring: 51-76. Rajan, Raghuram G. and Luigi Zingales (2003). Saving Capitalism From the Capitalists: Unleashing the Power of Financial Markets to Create Wealth and Spread Opportunity, New York: Crown Business. Ricketts, Martin (1999). The Many Ways of Governance: Perspectives on the control of the rm, London: The Social Aairs Unit. Rosen, S. (1981). The Economics of Superstars, American Economic Review, 71(5) December: 167-83. Scully, Gerald W. (2008). Sports. In David R. Henderson (ed.), Concise Encyclopedia of Economics, 2nd edn., available at http://www.econlib.org/library/Enc/Sports.html Vliamos, Spyros J. and Nickolaos G. Tzeremes (2012). Factors Inuencing Entrepreneurial Process and Firm Start-Ups: Evidence from Central Greece, Journal of the Knowledge Economy, 3(3) September: 250-64. Walker, Paul (2010). The (non)theory of the knowledge rm, Scottish Journal of Political Economy, 57(1) February: 1-32. Walker, Paul (forthcoming). The Reference Point Approach to the Theory of the Firm: An Introduction. Journal of Economic Surveys, available at http://onlinelibrary.wiley. com/doi/10.1111/j.1467-6419.2011.00714.x/abstract Ward, Benjamin (1958). The Firm in Illyria: Market Syndicalism, American Economic Review, 48(4) September: 566-89. Williamson, Oliver E. (1985). The Economic Institutions of Capitalism, New York: The Free Press.

35