On the Prediction of the Postal Service’s Marginal and Total Operating Costs

This chapter proposes an econometric framework aimed at predicting the United States Postal Service’s (the Postal Service’s) marginal and overall accrued (total operating) costs. The estimation is in part non-parametric in the sense that some of the parameters of interest are functional. It combines annual product-level data over the period 2000–2021 with aggregate annual data over the same period. The framework presented is suitable for simulating marginal and operating costs based on assumptions regarding volumes and network variables. The chapter adopts a reduced-form approach in the following sense: (i) The explanatory variables appearing in the equations are assumed to be exogenously determined while the dependent variables are assumed endogenous. (ii) The structures underlying these equations are more complex than reflected in the equations. This approach is justified in part by the fact that certain cost variables, e.g., marginal cost, which are part of the data underlying the estimation of the proposed econometric model, are themselves obtained from the econometric estimations that are performed upstream by the Postal Service (to causally link costs to their drivers while controlling for many relevant factors). So, it is important to caution that although the estimated links are suitable for making predictions and performing simulations, they are not necessarily causal.