All economic studies require high-quality data on costs to be meaningful. If the numerator of a CEA is measured with error, policy makers will lack confidence in the results. As a result, a CEA of the NMEP will be guided by the availability of cost data. Accounting data represent a recommended approach, if available. Cost models, or a hybrid of approaches, may be possible if accounting data are unavailable or poorly measured.
Accounting detail must be available at a sufficient level of detail to assign costs to one NMEP activity or another with no overlap or double counting. (For example, N MEP staff may work on non-NMEP activities or may share office space with non-NMEP activities. NMEP costs must be separated out from non-NMEP costs.) Data on the proportion of costs attributable to specific activities may be available for shared resources; if not, it is apportioned based on assumption, which may be examined in sensitivity analyses. Despite some apparent advantages of accounting data, they may be incomplete due to errors in reporting or in misallocation of costs.
Alternatively, a modeling approach may be applied if accounting data are poor. As discussed earlier, one could build a model of total costs by identifying all fixed and variable costs related to NMEP activities and staff. Assumptions about staff, labor costs, materials, material costs, and so on are combined to estimate a total. This total can also be calibrated or adjusted to fit an expected total, such as a grand estimate from accounting data.
A major challenge for the NMEP, regardless of the approach to cost measurement, is that summary fiscal budgets for FY98 FY06 indicate that NMEP activity costs have changed substantially between budget years. Thus, studying a single fiscal year may cause a particular NMEP activity to be an outlier. Studying several years of data may help if effects or outcomes data are also available; averaging costs is another possibility, if the same can be done for effects or outcomes.