HP Labs established the first industrial experimental economics program. Research in this program ranges from basic scientific studies to business applications. HP is the only company that has married advanced experimental economics technology to business decision making.
Research Areas
Wind Tunnel for Business:
The only way to account for human behavior is to have human beings in the system. Experimental methods, with human subjects, are used to help HP business divisions design and test "good" contracts and policies in their distribution channels. This work has resulted in a series of experiments examining contract terms from return policies to minimum advertise price policies. These experiments have provided HP business divisions with invaluable information to make better policy decisions. This technology serves as a "wind tunnel" that allows business to "test fly" policies and contracts in the laboratory before trying it out in the real market place.
Besides experimental research, HP Labs also has an ongoing research project with Caltech studying the same issues with theoretical mechanism design analysis. This work potentially can revolutionize how business makes decisions about contracting.
Information Aggregation:
Another area of interests is information aggregation. From 1996 to 1999, in collaboration with Caltech, HP Labs conducted a series of field experiments within an HP business. Markets of Arrow-Debreu securities were used to forecast various business events such as sales of certain products. This research shows the potential of using economics mechanisms as forecasting tools. In 2001, a new information aggregation mechanism for small groups was developed. Laboratory experiments have shown that this new mechanism can outperform markets significantly when information is independent and private across individuals. We also have developed an extension to the mechanism that will identify and aggregate a mixture of private and public information.
Reputation Research: We studied the effects of information policies (reputation mechanisms) on an exchange economy. The experiment model is a standard induced value supply and demand environment with a discrete version of the double auction as the market institution. The key feature of this model is that contracts traded in the market are not binding. Players make independent decisions of whether they fulfill their end of the contract. In addition, players can make offers (via the double auction mechanism) to only a subset of the players. In this scenario, reputation is driven by moral hazard and information of past behavior. Three information policies were studied including one that resembles the eBay feedback mechanism. This work is the first of a sequence of planned experimental research projects that will study how reputation propagates in a decentralized system and whether it is feasible to set up a market for trading reputation assets.
Quantum Game Theory:
A new area of research is to apply quantum mechanics ideas to game theory. Quantum generalizations of conventional games broaden the range of available strategies, which can help improve outcomes for the participants. HP Labs has developed a quantum approach to the economically significant n-player public goods game. In spite of the large temptation to free ride on the efforts of others in the original game, the quantum mechanism is sufficient to give near optimal expected payoff. Future work involves extending this methodology to design a reasonable business model to sell digital content on the Internet, which has properties resembling those of public goods economics.
Experimental Economics Software Platform Development:
HP Labs has developed a flexible experimental economics software platform (MUMS) that allows rapid implementation of economics experiments. The MUMS platform has been online for several years and has been responsible for many different types of experiments. Currently, we are developing the next generation software that will allow us to conduct experiments seamlessly on the web.
Recently, UC Berkeley started to develop their experimental economics infrastructure. MUMS will be deployed as one of the experimental platform in Berkeley's lab. HP Labs and Berkeley are also in the planning stage of joint software development for experimental economics.
Contact:
Proj Mgr: Fereydoon Safai
Project Lead: Kay-Yut Chen
|
printable version
