What Can Laboratory Experiments Teach Us about Emissions Permit Market Design?

Article excerpt

The laboratory provides a test bed to inform many design choices for emissions permit markets. Experiments are sometimes strongly motivated and structured by specific theoretical models and predictions, but in other cases the experiment itself can be the model of the market and regulatory environment. We review specific experimental applications that address design issues for permit auction rules, permit expiration dates and banking, liability rules, and regulatory enforcement.

Key Words: cap-and-trade, auctions, liability, regulation, compliance, banking

Emissions trading has grown significantly in importance in the regulatory toolkit over the past few decades. The flexibility allowed by this approach, which is often referred to as cap-andtrade, is estimated to have saved billions of dollars as a key component of efforts in the United States to cut sulfur dioxide (SO2) emissions in half during the 1990s. Emissions trading is now the centerpiece of many serious efforts worldwide to reduce greenhouse gas emissions. Emissions permits convey rights that can be exchanged between firms, subject to some constraints and rules chosen by the regulator, and these trades lower the overall cost of meeting an environmental goal by equalizing the marginal abatement costs across sources (Montgomery 1972). But what market rules and constraints are necessary, and which harm the performance of the emissions trading system? Should the regulator rely on trading institutions to emerge naturally, even if this results in relatively decentralized markets with high transaction costs, or instead take steps to encourage or even sponsor centralized trading institutions? The principle of emissions trading is elegant and simple, but market performance can depend on many design factors. The devil is in the details. Fortunately, laboratory experiments can create real, simplified, and controlled markets to help answer important questions like these.

The laboratory is useful because all markets are influenced by similar economic forces. Laboratory markets are populated by profit-motivated human agents, just as markets in the field are, and with induced incentives and other controls they are actual market microeconomic systems (Smith 1982). Different forces are stronger and more complex in different situations, of course, but they can be isolated and studied with careful experimental designs. Few economists would dispute that theoretical models are useful for providing insight into the factors that influence how markets perform. A theoretical model is a simplified abstraction constructed to help the researcher understand some real-world phenomenon. It is not intended to mirror every detail. An experiment is also a simplified construction, and it is often closely guided by a theoretical model. But the experiment is typically closer to the "real world" field than the theory is, because it includes human decision makers who may be boundedly rational or have nonstandard preferences that influence market outcomes in important ways. Experiments can thus serve as a bridge between theory and the field markets where our interest ultimately lies, since they indicate whether and how predictions developed through theoretical reasoning can be applied to more complex field conditions.

This article reviews some laboratory experiments that have been used to inform specific aspects of emissions permit design. One approach has been to develop theoretical models of the relevant market institutions and rules, and then evaluate the predictive value of these theoretical models using experiments that are constructed on the domain of the theories. The next section presents some examples of this approach. Sometimes theory provides less direct guidance, but the experiments can still provide a useful "test bed" to evaluate the researchers' or regulators' intuitions even when they are not modeled formally. The experiment itself is the model of the emissions trading system, and the impact of different design features can be investigated within the experimental model (Bardsley et al. …