In 2004, an Oregon referendum passed Measure 37, a law requiring compensation for, or waiver of, land-use regulations that decrease property values. A key question is whether this form of compensation for regulatory takings achieves economic efficiency. This paper shows that it does not under a reasonable set of assumptions. Following the recent law-and-economics literature on regulatory takings, this paper uses a two-agent principal-agent model in which the government makes a regulatory decision, the property-owner chooses investment level and/or land use, and the outcome is evaluated from the perspective of social welfare. An efficient rule would provide incentives both for the government to make socially optimal regulatory decisions and for the property-owner to make socially optimal decisions regarding investment in and use of the land. The paper shows that a full-compensation rule similar to Measure 37 can induce efficient decisions under a relatively simple model. The paper then extends the model by introducing factors brought to light by the form of Measure 37's compensation scheme and Oregon's early experiences under the law, including budgetary limits and limits on taxation. In this extended model, the Measure 37 rule is shown to lead to inefficiency in the form of under-regulation.