Data Related Issues in Index Numbers

Abstract

The material covered in the book so far addressed the problem of finding a suitable method for the construction of price and quantity index numbers for spatial and temporal comparisons. Most statisticians who are involved in the actual construction of these index numbers are faced with more basic problems of non-availability and non-comparability of price and quantity data. These are issues relating to the sampling design and the resulting estimators and their standard errors. Most consumer price index numbers are affected by the seasonality of commodities entering the consumption basket. Some commodities are either not available during certain months of the year or only available in small quantities at a very high price. A similar problem in the context of intercountry comparisons is that some commodities with a given set of specifications may not be available for consumption in some countries. In such cases, we have an incomplete price matrix with missing prices. This problem is particularly severe when the countries involved are heterogenous and are at different stages of economic development. Variations in the quality of goods and services consumed in different regions within a country or across different countries or over two time periods pose a difficult problem for statisticians. In a philosophical vein, these data problems appear to be more daunting than a simple choice of a suitable formula.