Governmental Loss Function

Government necessarily provides inferior service because of the scope of which it attempts. This is unavoidable, but losses can be reduced through particularization or decentralization. The Taguchi loss function formalizes the insights of Edward I. Deming. It can be applied to any production system or to services like government. The goal is to minimize variance from a customer's expectation target. In the case of government there are, of course, multiple targets. You can view governmental service as a multi-dimensional set of services. Some dimensions intersect with voter/customer preferences, others are for all purposes infinitely distant. Tax loss function presumably has a quality target of some low number. Some services are of concern to some voters, and the problem for government is to minimize losses arising from voters who are a finite distance from particular quality targets. In other cases, though, voters do not care about services.

Governmental loss functions cannot rotate without changing distance (amount of loss) for some and increasing it for others. Losses to taxpayers are increased as gains to tax consumers are increased. Moreover, the constellation of services can be changed only at taxpayer expense. Preferences form market segments. Identification of segments is part of political processes. However, total losses can increase even with a winning voter coalition. This is possible because vastly increasing losses for a minority can be accompanied by modest increases for the majority.

Limiting the scope of government would limit the losses because preferences can be targeted to varying segments. Loss minimization means better democracy and better satisfaction of public preferences. Thus, there is a quality advantage to decentralization.

The major threat to quality in government services is the Olson/Stigler economic loss due to political pressure. As size increases asymmetric advantages to specific groups increase. This is because lobbying and organizational costs increase and because small losses in a particular lobbying episode may be insufficient to motivate resistance. Multiple small losses across a large geography and population accumulate, resulting in far larger losses than would be possible in a smaller population and geographic area. Thus, satisfaction with state and local services is likely to be higher than satisfaction with federal government services. Lack of awareness of federal government services is likely to be greater than lack of awareness of state and local services.


Decentralized blocks can make more rational decisions because incentives to participate are greater.