Distributed energy resources, such as rooftop solar photovoltaics (PV), are likely to comprise a substantial fraction of new generation capacity in the United States. However, forecasting technology adoption based on peoples willingness to pay (WTP) faces two major challenges: the stated-intention and omitted-variable biases. Previous solar adoption literature has neglected to address these two biases altogether. Here, we adopt a "parameterization + calibration" approach to address both biases and estimate customers' WTP for PV. After collecting survey data on respondents' WTP for adopting PV, we characterize its empirical cumulative density function using a gamma distribution. We further calibrate the gamma distribution parameters using a national distributed PV adoption simulation model, finding the parameters that produce the best fit between simulated and historic solar adoption. We then show that the calibrated gamma distribution improves the raw WTP data after correcting for the two biases. Finally, we use our optimally-calibrated WTP to forecast market demand for residential PV at the county-level of the United States in 2020. Improving estimates of customer willingness to pay has significant implications for policy directly, e.g. estimating the effect of a proposed policy on technology adoption, and other regulatory processes that use forecasting, e.g. integrated resource planning.