Location: 368A Heady Hall
Description: Steve Miller
"Estimating Heterogeneous Effects of Transferable Quota Programs Using Causal Trees"
The effects of public policies often vary across the firms, individuals, or locations that they target. To estimate how a policy can differ in its impact across groups, standard econometric approaches require assumptions about the form of heterogeneity. In this paper, I extend and apply a recent method (Athey and Imbens 2016) which uses machine learning techniques to uncover the form of heterogeneity without sacrificing causal interpretation. I first extend the approach to allow for effects to vary not only cross-sectionally, but also with both policy duration and calendar time. I then apply this technique to revisit the effectiveness of an increasingly common market-based natural resource policy: Individual Transferable Quotas (ITQs) in fisheries. Prior work (Costello, Gaines, and Lynham 2008) demonstrated that ITQs reduce the probability of fishery collapse on average, yet ITQs continue to receive significant push back, in part because ITQs are not uniformly effective. Using a global dataset containing fisheries yields, ITQ status, and biological with information on over 11,000 fisheries, I find considerable heterogeneity in the effects of ITQs. Effects depend on the pre-policy state of the resource, how long the policy has been in effect, biological characteristics of the resource, and interactions among those dimensions. These findings can help target the use of property rights in fisheries management and set expectations about when policy effects will appear for different resources. More broadly, these results demonstrate a new method with applicability across a range of natural resource and environmental policy questions.
Contact Person: Ivan Rudik