Collective Action with Incomplete Commitment: Experimental Evidence.

• Author: Schmitt, Pamela

Pamela Schmitt [*]

Kurtis Swope [*]

James Walkert [+]

Face-to-face communication is investigated as an efficiency-enhancing mechanism in experimental common-pool resource environments in which the appropriation activities of outside appropriators create coordination and monitoring problems for the communicating group. We identify three distinct problems that can threaten successful collective action. Outsiders respond strategically to reductions in appropriation by cooperating group members. Members of the communicating group deviate from agreements more frequently when monitoring is imperfect and over appropriation can be blamed on outsiders. Groups that are allowed to communicate anticipate the potential problems and have difficulty reaching agreements or committing to a specific appropriation rule.

1. Introduction

   A growing literature in the social sciences examines the ability of individuals in group settings to make credible commitments to a particular course of action when substantial temptations exist ex post to deviate from such commitments. In such situations, a social dilemma arises in which the incentives to deviate from agreements lead to outcomes that are suboptimal from the perspective of the group. Economic environments involving common-pool resources (CPRs) have the incentive properties of a social dilemma. CPRs are natural or man-made resources in which appropriation by one individual creates an external cost on other users. Exclusion of individuals from appropriating from such resources is generally infeasible or nontrivial from either a technological, constitutional, or economic perspective. Fisheries, grazing lands, and irrigation water are common examples of CPRs in the field.

   Contemporary political theory, as well as noncooperative game theory, suggests that nonbinding agreements reached in the absence of an external enforcer are unlikely to change individual behavior and improve group outcomes. This is in stark contrast to empirical evidence from the field and findings from collective action experiments that demonstrate that under some conditions, individuals can successfully organize themselves and adopt strategies that improve individual and group returns without an external enforcer (Ostrom 1990; Ostrom, Gardner, and Walker 1994). Previous experimental research has shown that the ability of appropriators to communicate face-to-face in CPR settings can be a powerful tool for enhancing cooperation. Some hypotheses forwarded to explain why communication increases the selection of cooperative strategies focus on (i) offering and extracting of promises, (ii) the reenforcement of prior normative orientations, and (iii) the development of a group identity.

   As a continuation of several previously conducted laboratory studies, this paper focuses on the ability of individuals in CPR settings to overcome the inefficiencies that arise in social dilemma situations through face-to-face communication. Participants in our experiments make decisions about the investment of an endowment of tokens between two markets. Market 1 yields a certain, private return while the return from tokens invested in Market 2 is dependent on both individual and aggregate token investments in that market. By design, investment incentives in Market 2 have the characteristics of appropriating from a CPR. Investment by one individual lowers the average return to investments by all investors in Market 2. Following the literature in resource economics, this is referred to as rent dissipation. Outcomes in the class of CPR games discussed here are evaluated in terms of the percentage of maximum rents captured from the CPR. [1]

   In the set of CPR experiments reported by Ostrom, Gardner, and Walker (1994), hereafter OGW, a series of 10 decision rounds with no communication was followed by a series of decision rounds with face-to-face communication between each decision round. The series varied according to whether subjects were endowed with 10 or 25 tokens. The OGW experiments, combined with prior experiments allowing no communication, led to several important conclusions. Investments in Market 2 with no communication are characterized by a pulsing pattern. Investments increase, leading to a reduction in rents, at which time investors reduce their investments and rents increase. This pattern reoccurs across decision rounds within an experiment, with a tendency for the variation across rounds to diminish as the experiment continues and to converge somewhat near an inefficient Nash equilibrium. Investment behavior is also affected by token endowments. Rents as a percentage of optimum are less in 25-token experiments than in 10-token ex periments. The most striking conclusion, however, is that subject groups with repeated opportunities to communicate before every decision round obtain average rents that are substantially above those obtained in experiments without communication (99% compared to 34% in 10-token endowment environments and 73% compared to 21% in 25-token endowment environments).

   We extend the research on individual and group decision making in experimental CPR settings by introducing three new decision environments. We examine the robustness of communication as an efficiency-enhancing mechanism by creating environments in which the investment activities of outside investors create coordination and monitoring problems for the communicating group. Specifically, members of the communicating group face uncertainty of two types: (i) uncertainty about how the outsiders will strategically respond to investment changes made by the communicating group and (ii) uncertainty about the extent to which members of the communicating group follow agreements because of incomplete information on individual investment levels. Common to all decision environments is that only six of a total of eight of the participants are given the opportunity to communicate. In the first decision environment, which we refer to as Protocol 1, the two outsiders are computerized players, whose investments are determined b y random draws conducted by the experimenters. In the second and third decision environments, Protocol 2 and Protocol 3, the remaining two participants are human subjects, who are simply isolated from the other six and make independent investment decisions. These two outsiders are located in a separate room and have no verbal contact with each other or the other participants during the game. The difference between Protocols 2 and 3 relates to restrictions on Market 2 investments imposed on the outsiders.

   The results of our experiments provide additional insight into the prospects of self-governance of some commons as a viable alternative to privatization and government regulation. In particular, the experimental results expose the limitations of communication as an efficiency-enhancing tool in some CPR settings. [2]

   The paper is organized as follows. Section 2 describes the experimental CPR decision environment including experimental design, parameters, and theoretical predictions. Design details and experimental results from Protocols 1, 2, and 3 are presented in section 3. Section 4 compares results across the three protocols. Section 5 contains concluding comments.

2. The Experimental CPR Decision Environment

   Theoretical Predictions

   The Single-Play CPR Game

   The environment used in our laboratory decision situations can be interpreted as a limited-access CPR. [3] The class of single-play CPR games assumes a fixed number, n, of appropriators with access to the CPR. Each appropriator, i, has an endowment of resources, e, which can be invested either in the CPR or in an outside activity with a constant marginal return, w. The payoff to an individual appropriator from investing in the CPR depends on aggregate group investments in the CPR and on the appropriator's investment as a percentage of the aggregate.

   Let [x.sub.i] denote appropriator i's investment in the CPR, where 0 [leq] [x.sub.i] [leq] e. The group return to investment in the CPR is given by the production function F([Sigma] [x.sub.i]), where F is a concave function, with F(0) = 0, F'(0) [greater than] w, and F'(ne) [less than] 0. Initially, investing in the CPR pays better than the opportunity cost [F'(0) [greater than] w], but at some level' of investment ([x.sub.h]) the outcome is counterproductive [F'([x.sub.h]) [less than] 0]. Thus, concavity allows the CPR to yield maximum rents only when individuals invest some, but not all, of their endowments in the CPR.

   Let x = ([x.sub.1],[ldots],[x.sub.n]), be a vector of individual appropriators' investment in the CPR. The payoff to an appropriator, [u.sub.i](x), is given by

   we if [x.sub.i] = 0

   w(e - [x.sub.i]) + ([x.sub.i]/[sum] [x.sub.i])F([sum] [x.sub.i]) if [x.sub.i] [greater than] 0. (1)

   From Equation 1, if an appropriator invests his entire endowment in the outside alternative, he gets a sure payoff (we), whereas if he invests in the CPR, he gets a sure payoff w(e - [x.sub.i], plus a payoff from the CPR. An appropriator's payoff from the CPR depends on the yield from total investments, F([Sigma] [x.sub.i]), multiplied by his share of overall group investments ([x.sub.i]/[Sigma] [x.sub.i]).

   Since our experimental design is symmetric, there is a symmetric Nash equilibrium, with each player investing [[x.sup.*].sub.i] in the CPR. To see this, let the payoffs in Equation 1 be the payoff functions in a symmetric, noncooperative game. Then each player allocates [[x.sup.*].sub.i] in the CPR such that

   -w + (1/n)F'([[nx.sup.*].sub.i]) + F([[nx.sup.*].sub.i])[(n - 1)/[n.sup.2][[x.sup.*].sub.i]] = 0. (2)

   Compare this equilibrium to the optimal solution. Summing across individual payoffs [u.sub.i](x) for all appropriators i, the group payoff function u(x) is

   u(x) = wne - w [sum] [x.sub.i] + F([sum] [x.sub.i]), (3)

   which is maximized subject to the constraints 0 [leq] [x.sub.i] [leq] ne. Given the above conditions on F, the group...