Field Experiments and Environmental Economics A typology Harrison

Field Experiments and Environmental Economics

A typology (Harrison and List, 2005) Subject pool Framing Environment Conventional lab Students Abstract Imposed Artefactual field Nonstudents Abstract Imposed Framed field Nonstudents Commodity Imposed Natural field Nonstudents Commodity Natural So two important aspects of a natural field experiment: -Non student sample -“They do not know that they are being studied” and the act in an environment they naturally undertake the task.

Why do experiments at all? Can deal with a number of problems that we have with RP and SP data including: -Problems with endogeneity -Problems with correlation among variables -Problems with causality -Problems with hypothetical bias Historically, experiments were made in the lab. To test theories, for example the perfect competition model. But there is now a strong movement towards field experiments. Mainly in development economics, but also in behavioral economics. But very few in environmental economics (despite the frequent use of lab experiments)

Examples of natural field experiment questions - Does increased auditing reduce corruption? - Do textbooks at schools increase academic test scores? -Do people donate more money if they know how much others have contributed? The last one (charitable giving) is of interest for environmental economics, and perhaps the one most studied with field experiment. However, how this transfers over to contribution to “real” public goods is not clear. So for environmental economics: -Contribution to public goods -Use of common property resources -Effects and effectiveness of regulations, policies such as information and labelling. -Valuation?

Problems with field expeiment Main problem Ethical issues? Can we change the conditions in which public goods related to health and the environment are provided? - how useful is ex-post information about program effectiveness and can we generalize our findings? - more difficult to control - cannot always generate the variations that we want - expensive

An example of a social dilemma situation in the field Rural Vietnam • Goverment is in charge of larger scales of public good • Voluntary contribution in 2005 by villagers a wooden bridge. But it is now highly degraded. • Can they get togther and build a concrete bridge?

What we learned from the field -They village is in need of a new bridge - Last time they built the bridge it was funded by voluntary contributions of money and time. -Building a new concrete bridge would be expensive for the village

Our experiment -We collaborated with an NGO and provided money sufficient enough to build a concrete bridge. - But we wanted to learn more about factors that can affect peoples contributions, so we designed a natural field experiment. - In particular we want to study the role of social influence: o. Social information effects/conditional cooperation o. Default option effects

Social information and defaults Social information and voluntary contribution Lab experiments with public good games (Keser and van Winden 2000, Fischbacher et al 2001; Bardsley and Sausgruber, 2005) Field experiment with charitable giving (List and Lucking-Reiley 2002; Frey and Meier 2004; Landry et al 2006; Alpizar et al 2008; Shang and Croson 2009) Default options and choices: Empitical studies on pension savings (Crinqvist and Thaler 2004), car insurance (Johnson et al 1993), pro-social behavior (Johnson and Goldstein 2003)

The experiment: Threshold PG game The threshold public good experiment with refund policy and proportional rebate rule Succeed in providing the public good Fail in providing the public good oxi: contribution amount of household i o. Gi: marginal benefit from the public good o. N = 200 households o. Endowment E = 400. 000 VND (~ $23) o. Threshold T= 40 million VND o. Nash equilibria (examples): Each contributes zero Each contributes at the threshold level (T/N = 200. 000 VND)

The experiment: explaining to the subjects Each family has 400, 000 dong provided by the project. We would like to ask how your family would want to use this money. You can choose any amount to allocate to construction of the bridge. The concrete bridge will be established if all families contribute in total 40 million dong or more. This means that if the total contribution is equal or above 40 million dong, the project will use this money, add more funding in order to meets the costs of the bridge and take the responsibility to build up the bridge. If the total amount of money collected is more than 40 million dong, the excess amount will be returned to your family according to the proportion you contribute to the concrete bridge. If the families are unable to contribute in total 40 million dong, your contribution will be returned to you, and the concrete bridge will not be built. Your family is under absolutely no obligation to contribute any money to the concrete bridge, as these monies are yours. Even when your family is not willing to contribute or just willing to contribute a small amount, and if the bridge goes on operation, you have the full right of using the bridge because the bridge is a common property. No one in the commune, not even the officials, will know about your decision. We will keep your contribution information in secret.

The experiment: treatments An experiment and a follow-up survey (1 month later) 5 anonymous treatments: o. No social information and no default option o. High social information at 300, 000 VND and no default option o. Low social information at 100, 000 VND and no default option o. No social information and default option at zero contribution o. No social information and default option at full contribution Social information: ”typical previous contribution of others” Default option: a metal card with a magnetic token

The experiment: design in the field The experiment format: o. Experiment at home o 15 solicitors o 1 day experiment (Saturday afternoon and Sunday morning)

The experiment: challenges o. A realistic and credible experiment? →Pre-experiment tasks and information →Contract o. Size of endowment and threshold? → Study on WTP in another village o. Information leakage? →Saturday on one side of the bridge, Sunday the other side → 15 solicitors o. Solicitor biases? → Well trained and experienced

Results: raw data

Control variables Variables Definition Mean Household size Number of household members 3. 84 Age in year 49. 0 Education 1 = No schooling; 2 = Grade 1 -5, 3 = Grade 6 – 9, 4 = 2. 47 Grade 10 – 12; 5 = Vocational school Monthly income Household monthly income in hundred thousand 18. 13 Use the bridge 1 = use the bridge at least twice a month 0. 45 Cost of social events Monthly expenditures for “social events” e. g. wedding, and 196 different kinds of ceremony Member of party = 1 if member of the communist party 0. 10 Association = 1 if member of a local association 0. 49 Punish How likely is it that people who do not participate in 2. 59 community activities will be criticized or sanctioned? =1 very likely …=5 very unlikely Rice land Size of rice lands that are currently reaping in congs (1 4. 54 cong = 1/10 hectare) Saturday = 1 if experiment conducted on Saturday afternoon 0. 57

Results: regression Variable High social information Low social information Coeff. -11. 510 Coeff. 5. 488 -61. 378** -53. 325** Default at full contribution -12. 041 -2. 247 Default at zero contribution -48. 795* -51. 958** Household size Age 3. 685 0. 059 1. 860 -0. 349 Education 8. 918 -5. 370 Income 0. 304 0. 658 1. 125 27. 092 12. 969 0. 172** 0. 264 46. 072* 18. 984 0. 217*** 71. 237*** 43. 721*** -15. 408 -17. 494 Rice land Communist party member Association Social events Use the bridge Day of experiment Punish -19. 725*** Guess others contribution 0. 718*** No guess of others contribution Constant 149. 388*** 129. 616 104. 618

Summary and lessons learnt A field public good experiment: o. Contribution to a real public good o. Characterised by heterogeneity Significant and substantial effects o. Low social information: contribute 22% less compared to no information o. Zero contribution default option: contribute 20% less compared to average contribution. It was possible to implement the natural field experiment, although some problems were difficult to solve, and compromises had to be made.