The use of long-term contracts in the procurement of coal for electricity generation is common. The data that is observed from contracts and their transactions are from different levels of the pricing process. Contracts contain the parameters by which all future deliveries are structured, specifying the length of the agreement and acceptable coal attributes. Based on these parameters, a price is later determined for successive coal deliveries and the transaction occurs. This data structure fits well into multi-level models, where each level of the process is empirically estimated. A random intercept model is estimated where the first level is a hedonic model of coal prices. The contract that initiates the delivery is used to connect the two levels of the model. In the second level, contract coefficients from the first level are regressed on contract parameters to determine their impact on how coal is priced. Results find that many contract parameters are statistically significant in the price of coa l.
