This chapter provides an overview of the method of preparing the weighting diagram and a comparison of the weighting diagram of the revised series with those of the earlier series i.e., 1981-82 base series and 1970-71 base series. This has been done at the major group and group levels. The details of methodology, data and actual values of weights down to the individual item level and the related commentary on the comparative picture are presented in Annexure-III.

6.2 The major point to note in this exercise is the need to ensure representability of the overall price movements in the economy including those at disaggregated levels by a relatively small number of items and quotations. The major challenge therefore is to try and secure a sample that tends to represent the underlying universe adequately. In this exercise, it is also important to keep in mind that the weights would carry over to much of the next decade and thus a ‘futuristic’ outlook on the weighting diagram can not be avoided. Efforts, therefore, were made to see that certain items, such as hardware for the information technology industries, which were not so important in the base year and whose data were more elusive then than they are now, were included in the weighting diagram. It is also important to note in this exercise that the sample that is constructed is a one-shot opportunity and that the scope for "repeated samples" does not exist. Thus the characteristic of ‘unbiasedness’ of the sample had to be secured by design at the outset.

6.3 While the choice of the Laspeyre’s index provides the base ‘quantity’ weight, there is scope for a great variety of samples. As explained in chapter-4, the choice of the base year (1993-94) was made upon a deliberate search for a normal year of the economy which is also not very distant from the years of ‘data realisation’ (evolution of the series over time). It was also seen that the year fell within the phase in which a great deal of change has begun to occur in the economy with an array of new products being available.

6.4 Within the chosen base year, a myriad possibilities exist for the choice of the sample basket for the weighting diagram. The actual items chosen were dictated by the recommendations of the specific sub-groups appointed for the purpose. Their recommendations were scrutinised by the Working Group in terms of various parameters that were set and also from practical points of view. But ultimately the choice of items was constrained by the availability of data on a voluntary basis, on a continuous weekly basis starting with the week, 1 April 1993. Upon selection of the sample basket, the issue of distribution of weights to the various items assumes central importance.

6.5 There are two broad approaches to building up the weighting diagram. One method of weight distribution is to build it up "from below" i.e., add up the shares of each item within a group, then group shares into major group shares and then add up to arrive at "all commodities" normalised to 100. This method would maintain proportionality in the sample basket on the basis of their sample values. It is a moot point whether this method "from below" would provide an adequately unbiased sample to represent the underlying universe. In practice, there is no sure-fire test for this characteristic of the sample. It is very likely that the sample would deviate from the universe of all transactions in the economy. This is true for two reasons. There would be many missing values in the sample which may not follow the same commodity-wise proportionality that the census figures display. And secondly, they include all sectors of the economy including services. This second factor could be controlled, but the problem of comparability between the sample and the universe would be endemic except by fluke.

6.6 The other method is to construct the weighting diagram "from above". By this method, weights for the major groups would be imposed from outside and, depending on the sample concerned, weights within them at the group level could also be parametrically imposed. This method would ensure stratified distribution of weights, with shares at each stratum being defined rigidly. It is presumed that the scope for non-representativeness of the sample data within the confines of the stratum would be quite limited. This would then tend to throw up an adequate level of unbiasedness for the base weighted index. This method controls for the first factor imparting the bias in the method of construction "from below".

6.7 In the event, the top down approach was adopted. The services were excluded from the total value of transactions in the economy. The weights at the individual item level were generated from within the sample basket of the groups. From the major group "Manufactured products", the last group "Other Miscellaneous Manufactures" was removed due to anticipated sampling bias and its weight was distributed across other groups within the major group. The methodology of the bottom-up approach was also tested, but rejected by the Working Group because the resulting weights did not correspond with intuitive reasoning and hence gave rise to suspicions of possible sampling bias. The details of the exercises are discussed in Annexure III.

6.8 It turns out that the method of constructing the weighting diagram is similar in all major respects to the exercise carried out by the last Working Group.

6.9 The weights of the items are based on the value of transactions which consist of (a) value of (presumed) marketed surplus in the case of agricultural commodities and value of output in the case of non-agricultural products and (b) total value of imports less exports. Indirect taxes and duties were added as applicable. (A traded value threshold of Rs.120 Crore had been set as a parameter for manufactured products, but adjusted in some cases depending on exigencies, in the sample selection process). Within the agricultural sector, individual commodities have been assigned weights on the basis of average value of production during the triennium ending 1994-95, and the exact details are available in the report of the Sub-group on Agricultural Commodities. The marketed surplus ratios for the revised series and their corresponding marketed/marketable surplus ratios in the 1981-82 series are presented in Annexure II.

6.10 A comparative statement of the complete weighting diagrams of the revised series (1993-94 base) and the existing series (1981-82 base) of index numbers of wholesale prices in India upto item level is presented in Annexure III and upto group level in table 6.1. Graphs of the series upto the group level for the two series together with their 1970-71 values are presented below for sharper relief. The details of change are relegated to Annexure III, as these are mainly of technical interest.

*Weights for 1993-94 has been distributed over the groups (i - xii) of Manufactured Products.