Saturday, 8 August 2009

Just for fun: the theory of the firm 4

One of the standard approaches to the theory of the firm today is the property-rights view of the firm associated with the work of Grossman and Hart (1986), and Hart and Moore (1990). What is important in these papers is the ownership of, or (residual) control over, non-human assets. Such control generates the indirect influence that a "boss" has over his workers. Note that human capital is assumed to be inalienable. The influence the "boss" has, is due to him owning and controlling the physical assets which are important to the workers' productivity.

First up an obvious question is, What does the 'ownership' or 'control' of a physical asset entail? The Grossman/Hart/Moore approach borrows from the transaction-cost literature the ideas that contracts are typically incomplete: there are always unforeseen states of nature - or certain actions, such as making specific investments - that cannot be contracted upon. This in turn arises the question: Who decides how physical assets should be used in uncontracted-for eventualities? This is how ownership is defined. An owner of an asset has residual control rights, to use the asset in any way he sees fit except to the extent that particular usages have been specified in some initial contract. In particular, an owner of an asset has the right to deny access to anyone else.

This brings us to the issue of a definition of a firm. A firm is identified with the collection of physical assets over which the owner - the boss - has the residual control rights. Note that the boss exerts authority over workers because, in the event of a dispute, the boss can deny the worker access to important non-human assets. Moore (1992: 497) explains the importance of this,
I am going to show that if an agent does not own an asset, then his actions will depend on who does own it. For example, it will matter to the workers of a firm if their firm is taken over by another firm. The costs and benefits of integration can be understood primarily in terms of the (aggregate) effects on the incentives of the workers of the firms involved.
A simple example from Moore (1992) will help show this.
Suppose there is just one asset, a luxury yacht. There are three agents: agent 1, a chef; agent 2, a skipper; and agent 3, a tycoon. At date 1, agents 1 and 2 provide agent 3 with a service, gourmet seafare. We consider three models, with increasing degrees of complexity:

Model A. For the service to be useful, at date 0 the chef must take an action - say the preparation of a particular cuisine. This is a private effort decision, and cannot be contracted over. No other yachts are sailing nearby; hence the action is nontransferable. There are many other (potential) skippers at date 1; that is, agent 2 is dispensable. However, the tycoon is indispensable (only he can afford to fly to these waters). The cost to the chef of his action equals 100. The benefit to the tycoon equals 240. Finally, transactions costs prevent the writing of any long-term contracts at date 0.
So the question is, Will the chef take his action? Note that the action is socially efficient, since 240>100. The answer depends on who owns the yacht, the main non-human asset. If either the chef or the tycoon owns the yacht, then the date 1 (gross) surplus of 240 will be bargained over and split between the two of them, since each of them is crucial to the generation of the surplus (the chef has to provide the meal, and the tycoon is the only customer). Moore assumes that bargaining leads to both parties getting a half share, i.e., 120. The skipper receives none of the surplus, since he is dispensable. Since the chefs anticipated private return of 120 at date 1 exceeds his private cost of 100, he will take his action at date 0.

The other possible owner is the skipper. If he owns the yacht, then the surplus must be divided among all three of them (now the skipper is important, by virtue of his controlling access to the crucial asset). Here Moore assumes that three-way bargaining leads to each party getting a third, i.e., 80. Anticipating a private return of only 80 at date 1, the chef will not take his action at date 0 given that the cost is 100.

From this, one important point should be noted. The chef is more likely to take an action specific to the tycoon if the tycoon is his boss than if the skipper is his boss. A worker puts more weight on his boss's requirements than on someone else's. An employer indirectly gains 'authority' over an employee as a result of owning an important asset.
Model B. Modify Model A so that at date 0, the skipper also has an action-say to learn the history of the local islands to plan a better itinerary. Assume that the chef as well as the skipper is dispensable. The cost to the skipper of his action equals 100. The additional benefit to the tycoon equals 240. (So if both the chef and the skipper take their respective actions, there will be a total surplus of 480 to divide.)
In this case the question becomes, Will the chef, the skipper, or both take their respective action? Applying the same logic as in Model A, we can conclude:

Chef's share of his 240Chef act? Skipper's share
of his 240
Skipper act?
Chef owns yacht 120 Yes 80No
Skipper owns yacht 80No 120Yes
Tycoon owns yacht 120 Yes120Yes

The main result that can be seen from Model B is that when both the chef and skipper take actions specific to the tycoon, it is strictly better for the tycoon to own the yacht. That is, it may be efficient to give ownership of assets to agents who are indispensable even though they make no important effort or investment decisions.
Model C. As Model B, except that now the yacht comprises two pieces, the galley and the hull. These are strictly complementary, in that one is useless without the other. Also modify Model B so that the chef and skipper's actions are no longer specific to this tycoon; he too is dispensable. Finally, suppose that at date 0 the tycoon also has an action - say wooing business people to attend dinner parties aboard the yacht. The cost to agent i of his own action equals ci (i=1,2,3). The benefit of each agent's action to the tycoon equals 240 (making a total potential gross surplus of 720).
In this case Moore compares non-integration, that is, the chef owns the galley, and the skipper owns the hull with integration, that is, the chef owns the entire yacht (both galley and hull).

Applying the logic of Model A to this cases results in:
Chef acts iff Skipper acts iff Tycoon acts iff

Moore (1992: 499) writes,
For example, under nonintegration, for the tycoon to enjoy the additional 240 (gross) surplus arising from his own action, he must bargain with both the chef and the skipper since they each own a piece of the yacht. Accordingly, the tycoon's private return will be only 80. However, if the chef owned the entire yacht (integration), the tycoon would only need to bargain with one owner, and would obtain a private return of 120. Notice that the skipper's incentives to act are not diluted by losing control to the hull of the chef. The reason is that under nonintegration, the skipper has to bargain with the chef anyway (who owns the strictly complementary asset, the galley), and so it makes no difference to the skipper's incentives that the chef owns both pieces. The conclusion from Model C is that giving both pieces of the yacht to one agent (the chef in this case) leads to fewer hold-ups and greater efficiency. That is, assets that are strictly complementary should be owned together.
Importantly what Model C suggests to us is that complementary assets should be owned together, that is they should be integrated to form a single firm. A general proposition to this effect appears in Hart and Moore (1990: Proposition 8). There is a corresponding proposition, Hart and Moore (1990: Proposition 10), which tell us that asset that are economically independent should be owned separately owned, that is, they should not be part of a single firm. This is simply because from the viewpoint of the workers of the firm which has be bought, integration would amount to little more than bringing in an 'outside' owner who does nothing but dilute the incentives they face. This, then, means we have an explanation for the U-shaped average cost curve. Firms initially have increasing returns (stemming from the coordination of complementary assets) and then decreasing returns (stemming from the loss in incentives from outside control) - i.e. we have a theory for ,the optimal size of a firm based on transactions and technology.

Moore (1992: 500-1) makes a few comments on the theory developed above.
First, great emphasis is placed on ex ante (date 0) inefficiencies - i.e., underinvestment. We believe, however, that we may be picking up some of the same effects as one might obtain in a model in which there were ex post inefficiencies - due to the difficulty of striking bargains at date 1, particularly in a multilateral context such as this. For example, as in Model B, an agent who is crucial to the generation of ex post surplus (the tycoon) should have control rights, because agreement has to be reached with him anyway, so why increase the number of agreements necessary by giving control rights to others? It would be highly desirable to analyse the consequences of ex post inefficiencies in a systematic manner.

The next comment is related. The theory has no discussion of coordination, or information flows, at date 1. A boss never has to tell a worker what to do; the worker simply figures it out for himself. Equally, there is no discussion of hierarchy and delegation of decision making. A very important topic for future research is to extend the model to incorporate coordination, hierarchy and delegation.

Finally, the theory assumes that there is a date 0 market for assets which works perfectly, and that there are no wealth constraints. For example, in Model C, if the chef arrived at the date 0 market owning the galley, and the skipper arrived owning the hull, what would happen if the chef did not have the cash available to buy the hull? Remember, it is efficient for the chef to own both the galley and the hull,' so it is tempting simply to say that the chef borrows the money from a bank. But what security can the bank be offered in a world in which the chef has no resources except for his human capital - which, we have argued, is inalienably his?
  • Grossman, Sanford J. and Oliver D. Hart (1986). 'The Costs and Benefits of Ownership: A Theory of Vertical and Lateral Integration', Journal of Political Economy, 94(4) August: 691-719.
  • Hart, Oliver D. and John Moore (1990). 'Property Rights and the Nature of the Firm', Journal of Political Economy, 98(6) December: 1119-58.
  • Moore, John (1992). 'The Firm as a Collection of Assets', European Economic Review, 36(2-3) April: 493-507.

No comments: