Using survey-based choice experiments, we assess demand for alternative state-level carbon cap-and-trade programs. Household willingness to pay (WTP) depends upon more than just the emissions reductions that may be achieved. Households definitely care about changes in the numbers of carbon-intensive versus ‘green’ jobs in their own county and whether there will be additional regulations to limit non-global co-pollutant emissions from firms that buy permits. Preferences are somewhat less easy to discern with respect to the share of permits auctioned and revenue recycling options (i.e., the shares of auction revenue used to help households and firms to buy emission reducing equipment, or to support communities/workers that bear a relatively heavier burden due to the program’s costs). We specifically model response propensities among invited participants, which makes possible an ad~hoc correction for sample selection bias in the estimating sample, relative to the general population. Heterogeneous preferences are accommodated via mixed logit and latent class specifications, where individual respondent characteristics predict latent class membership. But we also estimate a model suitable for out-of-sample forecasting of WTP in other regions nationally, where systematic heterogeneity is captured by county-level characteristics. Main results include estimates of the marginal benefit exclusively from a one-tonne carbon emissions reduction (an alternative/complementary measure to the social cost of carbon), the marginal rate of substitution between carbon-sector and green-sector jobs, and a complete set of marginal willingness-to-pay estimates for each distinct program attribute. Our model with county-level observable heterogeneity uses (1) predicted county-level climate-change attitudes from the Yale Climate Map project, and (2) county-level COVID-19 case rates in prior months, given that the survey was fielded primarily during August and September of 2021. For selected carbon cap-and-trade programs, we conduct ‘benefits-function transfer’ exercises that reveal information about the predicted distributional effects of these programs. For each program scenario, outputs include the county-population-weighted marginal size distribution of predicted WTP across the lower-48 U.S. states, along with selected maps conveying the spatial distributions of predicted WTP across these 3,107 counties. The county-level predictions can also be aggregated to the state or regional level. Finally, we assess the differences in our model’s predictions with and without the effects of the COVID-19 pandemic, finding that predicted national mean WTP for carbon cap-and-trade programs would have been considerably higher had COVID-19 case rates been everywhere zero. (This paper evolved from one chapter in Garrett Stanford’s 2023 Ph.D. dissertation at the University of Oregon.)