Evaluating the Role of Regression Methods in the Determination of Standard Spending Assessments

A Standard Spending Assessment (SSA) is the amount which government considers appropriate for a local authority to set as its annual budget requirement. Each assessment is composed of revenues calculated for a discrete set of service blocks where the local unit costs of service provision are most often estimated by means of regression procedures that have been subject to regular scrutiny and review. Set against this context, in this paper the authors evaluate the performance of the regressions that contribute to the calculation of SSAs as equitable allocation devices. The general accounting framework for constructing SSAs is described and a statistical classification of the regression models is presented. In their analysis the authors examine a selection of models that are representative of each class and they include the regressions that comprise the Other District Services and Personal Social Services blocks. These appraisals follow a common sequence where a dummy variable is entered into the regression to test the consistency of the allocation made to a specified group of local authorities. When the dummy variable is statistically significant, revisions to the specification of the regression are either tested or suggested. Finally, the broader implications of these findings for the design of SSAs are examined.