Experimental Design
Many policies that experts recommend are not implemented by policymakers, because of their (ex-ante) unpopularity. Congestion charges are a case in point. This project aims at combining policy evaluation and the causal analysis of public support applied to the New York congestion charge and inform policymakers about two key items: how the congestion charge works (policy evaluation) and how voters learn about how it works and, as a result, possibly reconsider their stance vis-à-vis the policy (causal analysis of public support).
With policy evaluation, our objective is to contribute to a relatively limited strand of research that examines the impact of congestion charges on key outcomes such as congestion, accidents, and local air pollution. In line with the existing literature, we use administrative data for policy evaluation purposes. However, we also use survey data, which give us much more granular information on a variety of possible margins of adjustment in commuting behavior following the implementation of the congestion charge. Hence, our objective is to provide extensive evidence on the functioning of the New York congestion charge, a policy that affects the largest metropolitan area in the United States and one of the largest cities in the world.
With the causal analysis of public support, our objective is to provide evidence on how beliefs about and public support for congestion charges may evolve with direct experience of the policy.
This project also tests the extent to which information may be a substitute for experience. Information may also be a complement to experience, to the extent that the effects of a policy are not entirely salient to voters (with respect to the counterfactual). Hence, we are also interested in the extent to which information (provided at baseline) may be a complement to information.
To test the role of substitutability or complementarity of information with respect to experience, the baseline survey includes a randomized informational treatment, which describes the experience of frontrunner cities with congestion charges.
The survey covers respondents in the New York metropolitan area, respondents in four other metropolitan areas in the United States which currently do not have a congestion charge and have no immediate plans to introduce one, as well as respondents in the Greater London and in Singapore, both of which already have a congestion charge and so whose residents’ beliefs about congestion charges may be less affected by New York's experience with congestion charges. That is, the survey combines untreated control metropolitan areas and already-treated control metropolitan areas.
In short, the experimental design is as follows. Respondents are surveyed twice through two survey waves, to form a survey panel. At baseline, half of the respondents across metropolitan areas are exposed to the informational treatment. Between waves, respondents in New York's metropolitan area are exposed to the New York congestion charge. Hence, part of the sample is exposed to information only, part of the sample to experience only, part of the sample to both information and experience, and part of the sample to neither.