We develop an empirical methodology to study markets for services. These markets are typically organized as multi-attribute auctions in which buyers take into account seller's
price as well as various characteristics, including quality. Our identification and estimation strategies exploit observed buyers' and sellers' decisions to recover the distribution of sellers' qualities, the distribution of seller's costs conditional on quality, and the distribution of buyers' tastes. Our empirical results from the on-line market for programming services
confirm that quality plays an important role. We use our estimates to study the effect of licensing restrictions and to assess the loss of value from using standard rather than
multi-attribute auctions as is common in public procurement.