should they reflect expenditure or similar shares at the state level? The question of which weights to use should be examined further through future research.
While we feel obligated by our charge to raise the possibility of single-institution and state-level indices, the panel remains uncomfortable with the prospect that this might invite use of the model for accountability purposes. No such invitation is intended! Single-institution results may exhibit considerable volatility due to short-term variations in key input and output variables and the likelihood of data errors. Such volatility will likely decline as the number of institutions in the set increases. More importantly, as we have already emphasized, is the need to deploy robust quality assurance procedures in any situation where high-stakes quantitative productivity measures are used. In the slightly longer term, these quality assurance procedures should be supplemented by improvement-oriented structural models of the kind discussed in Chapter 2.
The first enhancement to the model is to track key labor categories separately from total FTEs. Separate tracking of labor types is typically a feature of sectoral productivity studies, but it is less commonly used to distinguish full-time from part-time employees; clearly, this differentiation is likely to be important in higher education. There are four reasons for this view.