A subjective model of experimentation

Typically an experiment is defined by a set of possible signals and a likelihood function, and both are specified exogenously—they are taken to be observable by the analyst. This paper renders them subjective by showing that they may be derived from suitable choice behavior. This is done in the context of an axiomatic representation theorem for preference on a suitable domain.     

Information and experimentation in short-term contracting

Summary. The impact of information dissemination and experimentation on dynamic adverse selection in noisy agency relationships is examined. Significant deviations in terms of equilibrium actions and payments occur, when compared to deterministic environments. Information dissipates slowly, so payments to agents who stand to lose informational rents over time are lower than compared to deterministic settings. Moreover, the principal manipulates the agent's actions to affect the informativeness of the signal. Thus, the principal trades-off lower initial payments with higher informational rents later. Simultaneously, the principal manipulates the signal distribution to enhance his ability to learn about the agent's type. Received: February 15, 2000; revised version: August 29, 2000     

Strategic experimentation and disruptive technological change

This paper studies the diffusion of a new technology that is brought to market while its potential is still uncertain. We consider a dynamic game in which an incumbent and a startup firm improve both a new and a rival old technology while learning about the relative potential of both technologies. The main findings are that (i) risk considerations make incumbents with higher market shares more likely to adopt the new technology and (ii) changes in market power are often preceded by a subpar performance of the new technology. We also show that introducing a better new technology or confronting a worse old technology may hurt the startup firm as its new technology is then adopted earlier by incumbents.     

Thought experimentation and the Phillips curve

This paper offers the rationale for presenting a particular type of Phillips curve and develops the dynamic behavior of an economy where such a Phillips curve relation is observed. The specific kind of relation that is explored has similarities with the sticky-information Phillips curve of the Mankiw-Reis framework. Nevertheless, it adds an important dimension: firms need to form expectations about current events on past time periods not because of infrequent optimal updating of information but because producers want to evaluate the possibility of taking advantage of information deficiencies on the consumers’ side. A positive probability of ‘fooling’ consumers with a price above the one imposed by market conditions re-shapes the dynamic relation between the inflation rate and the output gap.     

Virtual world experimentation: An exploratory study

We explore the scientific potential of virtual worlds for experimental economics in terms of the subject pools and experimental platforms they present. Our results offer tentative, qualified support for virtual world experimentation. Overall, the behaviour of virtual subjects recruited, incentivised and observed within Second Life across a range of five standard experimental games was not found to differ significantly from established standard results. In addition, we identify certain methodological opportunities and challenges which confront virtual world experimenters.     

Dynamic duopoly with learning through market experimentation

Summary This paper analyzes how learning behaviors can substantially modify the outcome of competition in an oligopolistic industry facing demand uncertainty. We consider the case of a symmetric duopoly game where firms have imperfect information about market demand and learn through observing the volume of their sales. The main body of the paper consists in showing how market experimentation can explain the existence ofprice-dispersion in an oligopolistic industry. We study this phenomenon and its dynamic evolution in the context of an Hotelling duopoly model; we then extend the analysis to general demand functions and toN-firm oligopolies. We discuss some implications of the public good aspect of information about market demand. We then conclude with a few comments on what happens when the value of information in the oligopolistic industry is negative.We are very grateful to Patrick Bolton for his helpful advice. We also wish to thank Richard Caves, Anthony Creane, Jean-Jacques Laffont, Andreu Mas-Colell, Eric Maskin, Jean-Charles Rochet, Margaret Slade, John Sutton, Jean Tirole, Mike Whinston and an anonymous referee.     

Holistic experimentation versus decomposition: an ultimatum experiment

In phase I, every participant plays the ultimatum game with each of the other five group members. For each of five offers, it is learned how many participants in each group would have accepted it. In phase II, the pie is 30 times larger. Thus, response behavior can be explored in phase I and, with this information, possibly exploited in phase II. Neither game theory nor equity theory suggest such holistic experimentation. Seventy-four participants out of 91 engaged in experimentation by submitting different offers in phase I. The remaining 17 submitted equal offers as suggested by decomposition hypothesis.     

Competitive experimentation with private information: The survivor's curse

We study a winner-take-all R&D race between two firms that are privately informed about the arrival rate of an invention. Over time, each firm only observes whether the opponent left the race or not. The equilibrium displays a strong herding effect, that we call a ‘survivor's curse.’ Unlike in the case of symmetric information, the two firms may quit the race (nearly) simultaneously even when their costs and benefits for research differ significantly.     

Parallel experimentation: a basic scheme for dynamic efficiency

Evolutionary economics often focuses on the comparison between economic competition and the process of natural selection to select the fitter members of a given population. But that neglects the other “half” of an evolutionary process, the mechanism for the generation of new possibilities that is key to dynamic efficiency. My topic is the process of parallel experimentation which I take to be a process of multiple experiments running concurrently with some form of common goal, with some semi-isolation between the experiments, with benchmarking comparisons made between the experiments, and with the “migration” of discoveries between experiments wherever possible to ratchet up the performance of the group. The thesis is that parallel experimentation is a fundamental dynamic efficiency scheme to enhance and accelerate variation, innovation, and learning in contexts of genuine uncertainty or known ignorance. Within evolutionary biology, this type of parallel experimentation scheme was developed in Sewall Wright’s shifting balance theory of evolution. It addressed the rather neglected topic of how a population on a low fitness peak might eventually be able to go “downhill” against selective pressures, traverse a valley of low fitness, and then ascend a higher fitness peak. The theme of parallel experimentation is used to recast and pull together dynamic and pluralistic theories in economics, political theory, philosophy of science, and social learning.     

Learning through experimentation in an oligopoly market with asymmetric information

Unlike previous literature, in which firms compete in the market with the same information, this article analyses a two‐period duopoly game in which only one firm is completely informed about the market conditions, whereas the other firm is unaware of one parameter of the demand curve. In this setting, we describe how the informed firm uses its price set in period 1 in order to reveal or to hide its private information and how the uninformed firm uses its own price in period 1 in order to learn the market conditions when they are not revealed by its rival. Specifically, we obtained the conditions under which the informed firm sets a higher price than its optimum in the first period to hide its private information in certain cases and to reveal that information in others. Likewise, this paper describes the conditions under which the uninformed firm sets a lower price than its optimum in period 1 in order to learn the unknown parameter. We found that the informed firm's cost of revealing its private information to its rival is lower than the uninformed firm's cost of learning the market conditions.     

The simple analytics of information and experimentation in dynamic agency

Summary. The dynamics of a stochastic, two–period principal–agent relationship is studied. The agent's type remains the same over time. Contracts are short term. The principal designs the second contract, taking the information available about the agent after the first period into account. Compared to deterministic environments significant changes emerge: First, fully separating contracts are optimal. Second, the principal has two opposing incentives when designing contracts: the principal ‘experiments,’ making signals more informative; yet dampens signals, thereby reducing up–front payments. As a result, ‘good’ agents' targets are ratcheted over time. Received: November 28, 2000; revised version: December 1, 2000     

The politics of experimentation: Political competition and randomized controlled trials

This paper provides an analysis of how political factors affect the incidence of the evaluation of public policies, with a focus on Randomized Control Trial (RCT) experiments in international development. We argue that political environments where incumbents face greater electoral competition and smaller ruling margins are more likely to host RCT experiments. Using various data sources for the incidence of RCTs both at the cross-country level and at the sub-national level in India, we find that RCTs are more likely to occur in politically competitive jurisdictions. We employ fixed effects regressions using various estimators and an instrumental variable strategy that exploits an electoral reform in India which limited the entry of independent candidates and exogenously affected the degree of electoral competition in state-level politics. The effect seems concentrated on RCTs that have the government as a partner, suggesting that political competition has an important demand-side effect on the incidence of RCTs.     

Duplication or experimentation: how local governors make policies under new administrative circumstances

This article shows that turnover of local governors increases the similarity of fiscal policies using the data of 320 prefectural-level cities from 2005 to 2012 in China. More generally, local governors duplicate successful experience rather than explore policies accommodating the specificity of their ruling areas when they are faced with new administrative circumstances. The structure and effects of the network resulting from the imitation of policies could be an important research topic in the future.     

Governance experimentation and factors of success in socio-technical transitions in the urban water sector

The necessity of a shift towards more sustainable urban water management practice is widely acknowledged and advocated. Experimentation that enables social learning is regarded of high importance for realising such a change. For instance, literature on Transition Management suggests that governance, as opposed to purely technical, experimentation is considered a critical factor in achieving a socio-technical transition. When analysing the water sector it becomes clear that modern urban water systems have almost exclusively focused upon technological experimentation with little attention directed towards the importance of governance experimentation for social learning. Empirically little is known neither on how governance experimentation actually unfolds nor about its effectiveness for socio-technical transitions. This research paper presents a critical analysis of a unique process of governance experimentation within the Australian urban water sector which generated sufficient social–political capital to change an established water governance framework. Conclusions of this research reveal some theoretically conjectured processes, like deepening, broadening and scaling-up, are found in this contemporary, real-life example. Furthermore, factors which influenced the success of this governance experimentation process are revealed and the role of various forms of learning therein is described.     

Biotechnology as a low-level-of-coherence policy priority: effectual-targeting and the need for learning and experimentation capabilities

During the last two decades, policymakers in various positions have been adopting a systems-approach to policy thinking. However, in contrast with the quasi-evolutionary way in which policy is thought of and conceptualised it seems that policy doing is still being guided by ‘market-failure’ justifications causing actual policies to remain narrow in their scope. In this paper we sidestep this analytical divide of thinking/doing by developing a co-evolutionary framework that utilises a more productive analytical divide of means/ends. That is to say we focus on a process of co-evolution of a higher-order, one which takes place ‘inside’ the agent of policy herself, and involves changes in the ways ‘means’ and ‘ends’ are understood and acted upon. Conceptualising policy problems in terms of means/ends contributes to current debates by rendering the difficulty that countries are facing in their attempts to prioritise biotechnology more intelligible and thus manageable.     

An experiment on learning with limited information: nonconvergence, experimentation cascades, and the advantage of being slow

We present the results of an experiment on learning in a continuous-time low-information setting. For a dominance solvable version of a Cournot oligopoly with differentiated products, we find little evidence of convergence to the Nash equilibrium. In an asynchronous setting, characterized by players updating their strategies at different frequencies, play tends toward the Stackelberg outcome which favors the slower player. Convergence is significantly more robust for a “serial cost sharing” game, which satisfies a stronger solution concept of overwhelmed solvability. As the number of players grows, this improved convergence tends to diminish, seemingly driven by frequent and highly structured experimentation by players leading to a cascading effect in which experimentation by one player induces experimentation by others. These results have implications both for traditional oligopoly competition and for a wide variety of strategic situations arising on the Internet.