Presents a study which uses data from management practice in a nonprofit budgetary setting to test whether input cost shares are independent prices while controlling for operating characteristics. Background on substitution among production inputs; Model for testing the fixed cost share hypothesis.

[Productivity measurement has not generally been considered part of the information that managers use in planning and control decisions. Kaplan (1983) criticizes accounting research for the lack of studies on production efficiency stating that the effects of output volume and substitution possibilities among key production inputs on productivity measures have not been the subject of any accounting research. Hasseldine (1967), Mensah (1982), and Marcinko and Petri (1984) have addressed the short-comings of traditional financial measures of productive efficiency. Barlev and Callen (1986) show that input standards should not be defined independent of input prices, the state of technology, and the level of output. These authors, however, provide no empirical applications. This study provides empirical evidence related to performance measures of efficiency of production. Traditional budgeting methods and measures used for analysis may provide inadequate information for effective performance evaluation and control monitoring. This is particularly true if the budget model assumes that input cost shares are fixed. Analytical methods including budget analysis that fail to consider available input substitution possibilities in response to changes in relative input prices, and methods that fail to consider changes in operating conditions, may result in lost opportunities for cost savings. For example, traditional methods of assigning responsibility for accounting variances tend to focus attention on meeting the budget and may divert attention from production input-mix decisions when relative input prices change. For this study, we obtained data on the output produced and the input consumed via on-site visits to 33 county correctional institutions (jails) in Tennessee. A multivariate regression system of simultaneous equations and a translog cost function specification were employed to analyze these data. Our empirical model specification required input prices, output levels, and the state of technology as independent variables (not summary financial measures) to explain total operating expenditures for the budget period. Agency theory provides a means for inferring managerial behavior. The translog cost function coefficients provide essential information about variability in input cost shares for the sample data. Based on the hypotheses tested, we rejected the reasonableness of the conventional budget model assumption of fixed cost shares. We provide empirical evidence that managerial decisions based on matching expenditures and appropriations in line item budgets may not be cost-minimizing. The evidence suggests that a translog budget model may produce additional useful performance evaluation and control monitoring information that is not available from budget models which assume that cost-minimizing input cost (budget) shares are fixed.]