- Ledyard1995 (ErikWJ): Ledyard, J. (1995). Public goods: A survey of experimental research. In J. H. Kagel & A. Roth (Eds.), The handbook of experimental economics (pp. 1-53 only - through section 3.3). Princeton, NJ: Princeton University Press. [available through course reserves∞]

A quick note: Page numbers in this write up are based on the page numbers in the article, not on the PDF document.

Reading Strategy

If for some reason a person does not have time to read all 56 pages of this paper I recommend that they read:

• The first 7 pages
• Section 1.5
• the introduction to section 2
• skim the section headings and section intros
• the “summary to this point” section on page 34
• skim section three paying attention to the headings
• read this review.

I want to be very clear that it is in your best interest to read the entire paper in great detail, but just in case….

Overview

“The research problem underlying this survey, then, is to understand behavior in the presence of public goods and in the context of many institutions…. On a broader level, the search is really for useful principles of behavior that apply across all environments and institutions. ” P3

“In this paper we will concentrate on… experiments with voluntary contributions mechanisms over a wide range of environments.” P. 10

For the general reader, there are two key points to take from this paper, what general design of public goods experiment look like. Secondly, that there three main findings that were yielded by examining six early experiments on experimental designs.

For the economist or person with a specific interest in the study of public goods, there is much more to this paper. For one it is a good survey of a series of experiments and how to write up a cross experiment analysis. It introduces the concept of mechanism design. It also provides a good analysis of early research into mechanism design and type of analyses that can be made on a wide range of experimental designs.

For all researchers, so most people in 884, there is also a good section on page 5 that shows that even a very simple experiment faces several major design choices that can each have effects on the outcome of the experiment. When not aware or controlling for each of these choices, they might unknowingly influence the results and lead to false conclusions.

Key Points/Claims/Methods

Simple example of a public good experiment

[paraphrased for simpliciy] Four people are brought to a room. They are each given $5. They are then told that each can choose to invest some or all of their $5 in a group project. In particular, each will simultaneously and without discussion put an amount between $0 and $5 in an envelope. The experimenter will collect the “contributions,” total them up, double the amount and then divide this money among the group. The private benefit from the public good, in this case, is one half the total contributions which is what each receives from the group project. No one, except the experimenter, knows others’ contributions but all know the total. The procedure is implemented and the subjects are paid. The data collected, beyond the description of the experimental parameters, is simply the amount contributed by each individual. p. 2

Definitions
PaulResnick: I think it's worth defining "public good" here. It's generally defined by two properties: non-exclusive and non-rival. Non-exclusive means that once it's produced, everyone gets the benefit (or harm in the case of a public bad, see below), regardless of whether or how much they contributed to it. Non-rival means that each person's utility for the public good is unaffected by other people's utility for it: use doesn't use it up.

The Language of Experiment: Mechanism and Environments

Mechanism design: a framework that includes environments, outcomes, performance criteria, institutions, and models of behavior. Each of these words is developed in greater detail in section 1.3, p. 6

[DerekHansen: Section 1.4 describes how Common Pool Resources, [discussed in Ostrom2000 (additional reading) and referred to heavily in this weeks reading Kollock1996] and Public Goods are inherently related. Economic theory suggests that Public Goods are underproduced, while Common Pool Resources (e.g., fishing grounds) are over-used. Ledyard makes the point in this section that Common Pool Resources are in essence no different than Public Goods. If I understand it correctly, you can just re-label the variables of a Common Pool Resource and end up with it looking like a Public Good problem. For example, if the Common Pool Resource is a fishing ground (where the incentives of individuals are to over-fish and thus deplete the public resource available to others), one could characterize it as a Public Goods problem by thinking of each person's promise to not excede their fish limit as their personal investment (t) toward the Public Good.

PaulResnick: Your understanding is correct, though it's not promises but actual actions that make the difference. The mathematical transformations let you think about the same problem in one of two ways. a) Think of increases in the fish stocks as a public good; the base amount is the fish stocks when everyone fishes as much as possible; restraint by individuals enables more production of the public good, which makes whatever fishing does happen more productive. b) Think of decreases in the fish stocks as a public "bad"; the base amount of the fish stocks occurs when no one fishes; fishing by individuals creates more of the public "bad", which makes whatever fishing does happen less productive.

The key idea with either public goods or common pool resources is that the actions of the individuals generate externalities, something that affects the utility of everyone, either a public good or a public bad.]

Are People Selfish or Cooperative?

“A reasonable reading of the literature on voluntary contribution mechanisms and social dilemmas would probably lead one to conclude that the major findings to date are:
1. In one-shot trials and in the initial stages of finitely repeated trials, subjects generally
provide contributions halfway between the Pareto-efficient level and the free riding level,
2. Contributions decline with repetition, and
3. Face to face communication improves the rate of contribution.” P 13.

(I expect that someone from Olson’s shapefactory group will have something to say about this, especially the communication points).

BenjaminChiao: In Olson's experiment, there is a finding of trade bias. The face-to-face group tends to trade mostly within the same group not with the group in another room. I guess one possible explanation is that the face-to-face group sees themselves in a subgame of a supergame that involves implicit payoffs in *other* interactions with the group members outside the experiment because the group membership is not anonymous. In the public good experiments, individual contribution, however, is almost always private information. If Olson's finding carries over to the public good experiment, will #3 above weaken as the group size increases even though the public good experiment is a face-to-face treatment simply because the difficulty of individual identifiability increases? [DerekHansen: Section 3.3.2 discusses the effects of group size on contributions to public goods, with some inconclusive but suggestive results. One argument is that higher group size will reduce contributions to public goods because non-cooperative behavior is harder to detect (as you suggest). An alternative argument is that your individual contribution will help more people if N is high (e.g., your marginal effect on social welfare), so if you are altruistic you will contribute more.]

YongKim: In the Shapefactory experiment, social presence and convenience are thought to be the primary factors driving the in-group trading bias of the FTF group. This finding was quite interesting in that the behavioral bias was simply created by the factor of 'being together' without any sort of sub-group identity development. Thus, it suggests how influential FTF can be on participants' behaviors.

Critique

Because we are trying to increase dialog on the wiki, I am going to post some of what I see are the controversial quotes from the paper in hopes that they will start conversation.

“Since both the economic/game-theoretic and socio-psychologic theoretical predictions are wrong, we need to discover more about behavior not only in the context of voluntary contributions but also in the presence of many institutional designs. Experiments are the only way to do so.” P3.

“The public goods environment is a very sensitive one with much that can affect outcomes but are difficult to control. The many factors interact with each other in unknown ways. Nothing is known for sure. Environments with public goods present a serious challenge even to skilled experimentalists and many opportunities for imaginative work. “ p 1.

“I believe that mechanism design and test bedding will ultimately become the foundation of policy analysis. P.11”

If the language of math is universal, why are there such variations of behavior among cultures and also, why do humans not find the nash equilibrium? Are there some cultures that try to economically optimize more than others?

BenjaminChiao: The concept of public goods is originated by Lindahl and generalized by Samuelson. There are various definitions of public goods but all of them contain the *non-excludability* property. See E. Lindahl, Die Gerechtigkeit der Besteuing. Lund: Gleerup.
[English translation: Just Taxation—A Positive Solution] (1919); P. A. Samuelson, Economics: An Introductory Analysis (1964). It has been widely accepted that it is efficient for governments to provide public goods for free [DerekHansen: or through taxation. PaulResnick: taxation is how funds are raised; for free refers to no charge to consumers]. However, to argue whether an economic good is excludable, we can not consider the good in isolation from its environment. As an illustration, should the government provide apples for free? Perhaps or perhaps not. The answer may be affirmative in an imaginary island in which the only possible means to defend invaders is apple (note: national security is a classic example of public good). But it will be a joke to provide them for free in the city of New York. From this perspective, one can argue that some private good is (or at least, in principle, can be) a national security end product. The series of theories and experiments surveyed in the paper, however, treat excludability of a good as exogenous. [DerekHansen: The experiments discussed in this paper are not really concerned with how to apply the principles to specific public goods situations (e.g., whether or not apples are public goods), but rather are interested in uncovering the most fundamental human behaviors related to public goods.] There never involves a belief elicited from the subjects about whether they think a good is excludable or not. [DerekHansen: This is because the rules of the experiment determine whether or not the good is excludable. This is why experiments are nice, the environment and institution (mechanism) can be set up as desired so that there is no ambiguity.] This, I conjecture, will change the set of public goods considered, and the corresponding aggregate contribution level given that the size of membership changes with exogeneity.

LaurieBuis: Just a comment on the quotes above as well as the rest of the paper... I get the sense that Ledyard sometimes forgets the ultimate struggle of the fundamental nature of social science. To date (and probably from here on to the future) we have never been able to predict human behavior with 100% certainty. This is a problem that in my opinion, we will never be able to overcome...

Language Critique

Economists speak a language that is all their own and they have spoke it for so long and trained so hard in it, they forget that others are not fluent at econ-talk.

For example take this paragraph:

The range of experiments which have a public goods structure is more extensive than most realize. To see why, let me describe some very simple environments with public goods. There are two goods, one private and one public, and N individuals. Each individual i = 1, ..,N is endowed with some amount of the private good, zi. The public good is produced from the private according to the production function y = g(t) where t is the amount of private good used to produce y. An outcome is a level of public good, y, and an allocation of the private good for each agent x1, ..., xN . Each agent values outcomes according to the utility function13 Ui(xi, y). Feasible outcomes are a = (y,x1, ..., xN ) such that y = g SN i=1(zi ?? xi) . We will call ti = zi ?? xi the amount of i’s payment for the public good and occasionally restrict the range of possible ti. For example, sometimes it is required that ti 2 [0, zi], the endowment is divisible but no one can contribute more than zi, nor can they repeat compensation, and sometimes it is required that ti 2 f0, zig, either zi is contributed or nothing is contributed. We can summarize the environment as e =< g,U1, . . . ,UN, z1, . . . , zN >. p9. (This section loses a bit in the conversion to the wiki)

LaurieBuis: I have to admit, I struggled with SI 502 and this brought back all of the old nightmares :)

Connections with other readings, ideas etc.

Axelrod: The Evolution of Cooperation

Olson & Olson: Distance Matters