Working Paper

Employment and Output Leakage under California’s Cap-and-Trade Program

May 13, 2016 | Wayne Gray, Joshua Linn, Richard D. Morgenstern


The possibility that economic activity may relocate from areas with high regulatory costs to ones with lower costs raises concerns about the potentially adverse impacts of a regional greenhouse gas cap-and-trade program.


To estimate the potential impact of California’s cap-and-trade program on the state’s energy- intensive, trade-exposed manufacturing industries, this paper uses confidential plant-level Census data to model the effect of historical energy prices on plant-level output, employment, and value added, both inside and outside California, holding constant foreign energy prices. Simulation of the model for an assumed compliance cost of $10 per metric ton of carbon dioxide equivalent (CO2) in California and zero outside the state yields 0 to 3 percent short-term (one year) impacts for almost a third of the industries studied with no output-based rebating. The largest losses are estimated in glass container manufacturing (17 percent), paperboard mills (14 percent), automobile manufacturing (13 percent), iron and steel mills and ferroalloy manufacturing (12 percent), and poultry processing (11 percent); these industries are among the most energy intensive of those studied. Estimated losses for another group of five industries are about 10 percent. These losses should be compared to an overall average one year loss of about 5.7 percent across all the California energy-intensive, trade-exposed industries studied. Simulations of higher compliance costs (up to $22 per metric ton of CO2) result in correspondingly larger losses. Over the long run, defined as a five-year period, the results suggest that increases in California's energy prices relative to those in nearby states have smaller effects than those effects seen over 1 year. Over this longer period, the largest output losses are below 1 percent, with most industries experiencing output losses below 0.1 percent, although for a variety of technical reasons the authors offer caution when interpreting the industry-specific long-run results.