During today's session of the Lowell Milken Institute for Business Law and Policy conference on competing theories of corporate governance, I will be presenting the above entitled mini-paper. It follows:
All organizations must have some mechanism for aggregating the preferences of the organization’s constituencies and converting them into collective decisions. As Kenneth Arrow explained in work that provided the foundation on which the director primacy model was constructed, such mechanisms fall out on a spectrum between “consensus” and “authority.” Consensus-based structures are designed to allow all of a firm’s stakeholders to participate in decision making. Authority-based decision-making structures are characterized by the existence of a central decision maker to whom all firm employees ultimately report and which is empowered to make decisions unilaterally without approval of other firm constituencies. Such structures are best suited for firms whose constituencies face information asymmetries and have differing interests. It is because the corporation demonstrably satisfies those conditions that vesting the power of fiat in a central decision maker—i.e., the board of directors—is the essential characteristic of its governance.
Shareholders have widely divergent interests and distinctly different access to information. To be sure, most shareholders invest in a corporation expecting financial gains, but once uncertainty is introduced shareholder opinions on which course will maximize share value are likely to vary widely. In addition, shareholder investment time horizons vary from short-term speculation to long-term buy-and-hold strategies, which in turn is likely to result in disagreements about corporate strategy. Likewise, shareholders in different tax brackets are likely to disagree about such matters as dividend policy, as are shareholders who disagree about the merits of allowing management to invest the firm’s free cash flow in new projects.
As to Arrow’s information condition, shareholders lack incentives to gather the information necessary to actively participate in decision making. A rational shareholder will expend the effort necessary to make informed decisions only if the expected benefits outweigh the costs of doing so. Given the length and complexity of corporate disclosure documents, the opportunity cost entailed in making informed decisions is both high and apparent. In contrast, the expected benefits of becoming informed are quite low, as most shareholders’ holdings are too small to have significant effect on the vote’s outcome. Accordingly, corporate shareholders are rationally apathetic.
In sum, it would be surprising if the modern public corporation’s governance arrangements attempted to make use of consensus-based decision making anywhere except perhaps within the central decision-making body at the apex of a branching hierarchy. Given the collective action problems inherent with such a large number of potential decision makers, the differing interests of shareholders, and their varying levels of knowledge about the firm, it is “cheaper and more efficient to transmit all the pieces of information to a central place” and to have the central office “make the collective choice and transmit it rather than retransmit all the information on which the decision is based.” Shareholders therefore will prefer to irrevocably delegate decision-making authority to some smaller group. As we have seen, that group is the board of directors.
Strong limits on shareholder control are essential if that optimal allocation of decision-making authority is to be protected. Any meaningful degree of shareholder control necessarily requires that shareholders review management decisions, step in when management performance falters, and effect a change in policy or personnel. Giving shareholders this power of review differs little from giving them the power to make management decisions in the first place. Even though shareholders probably would not micromanage portfolio corporations, vesting them with the power to review board decisions inevitably shifts some portion of the board’s authority to them. As Arrow explained:
Clearly, a sufficiently strict and continuous organ of [accountability] can easily amount to a denial of authority. If every decision of A is to be reviewed by B, then all we have really is a shift in the locus of authority from A to B and hence no solution to the original problem. 
This remains true even if only major decisions of A are reviewed by B. The separation of ownership and control mandated by U.S. corporate law effects thus has a strong efficiency justification.