A Social Contract Account for CSR as an Extended Model of Corporate Governance (II): Compliance, Reputation and Reciprocity

Abstract

This essay seeks to give a contractarian foundation to the concept of Corporate Social Responsibility (CSR), meant as an extended model of corporate governance of the firm. Whereas, justificatory issues have been discussed in a related paper (Sacconi, L.: 2006b, this journal), in this essay I focus on the implementation of and compliance with this normative model. The theory of reputation games, with reference to the basic game of trust, is introduced in order to make sense of self-regulation as a way to implement the social contract on the multi-fiduciary model of corporate governance. This affords understanding of why self-regulation, meant as mere recourse to a long-run strategy in a repeated trust game, fails. Two basic problems for the functioning of the reputation mechanism are examined: the cognitive fragility problem, and the motivational problem. As regards the cognitive fragilities of reputation (which result from the impact of unforeseen contingencies and from bounded rationality), the paper develops the logic and the structure that self-regulatory norms must satisfy if they are to serve as gap-filling tools with which to remedy cognitive limitations in the reputation mechanism. The motivation problem then arises from the possibility of sophisticated abuse by the firm. Developed in this case is an entirely new application of the theory of conformism-and-reciprocity-based preferences, the result of which is that the stakeholders refuse to acquiesce to sophisticated abuse on the part of the firm.

Appendix 1

This appendix sets out the explicit form of the overall utility function, which is expressed only in compressed form in the main text by the formula

$$V_i(\sigma)=U_i(\sigma)+\lambda_i F\left[{T(\sigma)}\right]$$

(A1)

where U _i is player i’s material utility for the state σ (a combination of individual strategies); λ _i is a weight that may be any positive (perhaps infinite) number; T is a fairness (to be specified) principle defined for the state σ; F is a function (to be specified) of the fairness principle expressing both the agent’s conditioned conformity and other individuals’ expected reciprocal conformity to T.

First to be specified is a form of the fairness-function T which represents the ideal formally. This must be a mapping from the set of states (and first-order utilities attached to them) to a fairness ordering ranging over states. A characterisation in contractarian terms of the ideal principle T is given by the Nash bargaining solution i.e. the N.S.W.F. $T(\sigma)=\mathop\prod\nolimits_{i=1}^N\left({U_i-d_i}\right)$ where d _i represents the reservation utility that agents can obtain when the bargaining process breaks down. In this case d _i coincides with a covering of the costs of each player’s specific investments, which means that fair bargaining on the surplus may only start if parties are assured that they will end up at least with reimbursement of the cost they must bear in order to participate in cooperation.

Then let us define the two personal indexes of conformity which are compounded in the measure F of mutual expected conformity and enter the utility function of the players. In this construction, I take the point of view of player i (any other player j’s perspective is symmetrical).

A. Player i’s personal index of conditional conformity

This is player i’s degree of deviation from the ideal principle T (which varies from 0 to −1) due to player i’s choice, given her expectation about player j’s behaviour. It is normalised by the magnitude of the difference between players' full conformity and no conformity at all, conditional on player j’s choice

$$f_i\left({\sigma_{ik}, b_i^1}\right)=\frac{T\left({\sigma _{ik}, b_i^1}\right)-T^{\rm MAX}\left({b_i^1}\right)}{T^{\rm MAX}\left({b_i^1}\right) -T^{\rm MIN}\left({b_i^1}\right)}$$

(A2)

where ${b_i^1}$ is player i’s belief concerning player j’s action, ${T^{\rm MAX}\left({b_i^1}\right)}$ is the maximum attainable by the function T given i’s belief, ${T^{\rm MIN}\left({b_i^1}\right)}$ is the minimum attainable by the function T given i’s belief, and ${T\left({\sigma _{ik},b_i^1}\right)}$ is the effective level attained by T when the player i adopts his strategy σ _k (where the index k means that player i’s strategy is chosen within a set where k may vary from 1 to N), given his belief about the other player’s behaviour.

B. Estimation of the second player index of reciprocal conformity

This is player j’s degree of deviation from the ideal principle T (which also varies from 0 to −1), as seen through player i’s beliefs – also normalised by the magnitude of the difference between player j’s full conformity and no conformity at all, given what j believes (and player i believes that he believes) about player i’s choice.

$$\tilde{f}_j\left({b_i^1, b_i^2}\right)=\frac{T\left({b_i^1, b_i^2}\right)-T^{\rm MAX}\left({b_i^2}\right)}{T^{\rm MAX}\left({b_i^2} \right)-T^{\rm MIN}\left({b_i^2}\right)}$$

(A3)

where ${b_i^1}$ is player i’s first-order belief about player j’s action (i.e. formally identical to a strategy of player j), ${b_i^2}$ is player i’s second-order belief concerning player j’s belief about the action adopted by player i (i.e. formally identical to a player i strategy predicted by player j).

These indexes are used to construct the following ideal component of the utility function

$$ \lambda_i\left[{1+\tilde{f}_j\left({b_i^2, b_i^1}\right)} \right]\Big[{1+f_i\left({\sigma_{ii}, b_i^1}\right)}\Big] $$

(A4)

where the weight λ _i is an exogenous psychological parameter that expresses, prior to any consideration of reciprocity, the extent of the disposition to act according to conformist considerations within the motivational system of player i. The formula states the following: if player i perfectly conforms with the ideal, given her expectation, while player j is also expected to perfectly conform, then the two individual indexes take value zero, so that the resulting utility value due to conformism is (1)(1) λ _i . Thus the maximum conformist utility value is λ _i . By contrast, if a player does not entirely conform, while not expecting the other player entirely to conform either, then the two indexes take negative values (possibly −1). Thus the utility calculation for conformist reasons reduces to (1−x) (1−y) (possibly both equal to zero) multiplied by the weight λ _i and gives less than λ _i (possibly zero) as the conformist utility value.

The overall utility function V _i is the linear combination of the two components

$$ \eqalign{V_i\left({\sigma_i, b_i^1, b_i^2}\right)= U_i\left({\sigma_i, b_i^1} \right)\cr +\lambda_i\left[{1+\tilde{f}_j\left({b_i^2, b_i^1}\right)} \right]\Big[{1+f_i\left({\sigma_{ii}, b_i^1}\right)}\Big]} $$

(A5)

This suggests that if a player predicts reciprocal conformism (as it enters the utility function), in so far as weight λ _i is high, it is then possible that the overall utility of a strategy choice reverses the effect of player i’s simple consequentialist preferences represented by U _i (σ _i , b _i ). For example, it may induce the player to select strategies that he would never choose if he relied on material utility only.

When overall utility functions are employed in game theoretical contexts, they require appropriate definition of the players’ best-response choices. Grimalda and Sacconi 2005 and Sacconi and Grimalda (2006) elaborate on Rabin (1993) in order to define a new model of reciprocity. Hence, as for Rabin, inclusion of beliefs in the arguments of the utility functions calls for extension from the standard concept of the Nash equilibrium to that of the (PNE) as defined by Genakoplos et al. (1989). The idea behind PNE is that, in equilibrium, the beliefs of rational players must be coherent with the strategies that are being played: that is, beliefs of any level predict lower level beliefs and actions. Hence also the result of fourth section is given in terms of the Psychological Nash Equilibria of the relevant game.