Introduction

In preceding chapters we have examined in detail a measure that, in principle, is superior to the various other indicators that form the basis of conventional applied welfare analysis. For the theorist, that examination should be sufficient. However, those concerned with actual use of operational procedures are bound to raise additional questions. Are the numerical procedures being proposed of sufficiently greater accuracy to warrant the scrapping of the traditional tools of analysis? Are the new procedures more complex? If so, does their greater accuracy outweigh their additional complexity and warrant abandonment of the somewhat less accurate measures currently in use?

In regard to the first question, it is impossible to establish any meaningful general criteria to determine whether or not the errors of approximation inherent in traditional approaches are likely to be small for the new procedures. Such errors inevitably depend on a variety of factors: (a) the initial price/quantity situation, (b) the magnitude of the variable change being evaluated, and (c) the characteristics of the approximation procedure being used. Thus, depending on the situation, it is likely that one particular method will work well in some circumstances but not in others and that performance will vary from situation to situation. This proposition must also inevitably hold for the approximation procedures based on the money-metric discussed in Chapter 3. There it was suggested that accurate measures can be constructed from information about the derivatives of the demand function up to the third order.