Introduction

The theoretical heritage of cost-of-living indices derives from two related lines of thought. On the one hand, Irving Fisher, in 1911, and again in 1922, proposed a number of criteria that such indices should meet in order to be considered “ideal.” The approach adopted can be characterized as being mechanistic and only very indirectly based on any consistent theory of consumer behavior [Frisch (1936) called it “atomistic”]. In contrast, the Russian economist Konös wrote in 1924 that when computing the true cost of living we should compare “the monetary cost of two different combinations of goods which are connected by the condition that during consumption of these two combinations, the general status of want-satisfaction (the standard of living) is the same” (1939, p. 10). This definition has subsequently been adopted by most of the leading writers in this area of investigation, such as Bowley (1928), Frisch (1936), Schultz (1939), Klein and Rubin (1947–8), and Malmquist (1953), and more recently Afriat (1972), Pollak (1971), Samuelson and Swamy (1974), Diewert (1980), and Sen (1979).

It is immediately apparent that the Konös index is based on the concept of the compensating variation, which we have shown in previous chapters to be unsatisfactory as a welfare indicator. Indeed, many economists have long been aware that something was amiss. Beginning in the 1930s, several investigators began to focus their attention on whether or not the Konös definition fulfilled the various Fisher test criteria. Unfortunately, the results have not been entirely satisfactory.