ISSN (Print) - 0012-9976 | ISSN (Online) - 2349-8846

A+| A| A-

Price Indices and Inflation Rates

With inflation shooting up, it is an appropriate time to critically discuss the quality of India's inflation statistics. This article first discusses the official wholesale and consumer price indices and the impact of not regularly incorporating new products and changes in quality, and related deficiencies. It also draws on an official United States commission report that studied the implications of productivity changes for measuring inflation. The paper concludes with suggestions for revising the procedures for collecting and compiling price information and also recommends preparation of a new producer price index to replace the official Wholesale Price Index and, in addition to the consumer price indices, the compilation of an annual "cost of living index".

SPECIAL ARTICLEEconomic & Political Weekly EPW June 28, 2008217Price Indices andInflation RatesT N SrinivasanT N Srinivasan ( at the Yale University, Connecticut, US.With inflation shooting up, it is an appropriate time to critically discuss the quality of India’s inflation statistics. This article first discusses the official wholesale and consumer price indices and the impact of not regularly incorporating new products and changes in quality, and related deficiencies. It also draws on an official United States commission report that studied the implications of productivity changes for measuring inflation. The paper concludes with suggestions for revising the procedures for collecting and compiling price information and also recommends preparation of a new producer price index to replace the official Wholesale Price Index and, in addition to the consumer price indices, the compilation of an annual “cost of living index”.The current spike in inflation has drawn attention to our present procedure of measuring inflation on a point-to-point annual basis using the Wholesale Price Index (WPI) with no seasonal adjustment. The governor of the Reserve Bank of India (RBI), Venugopal Reddy, has said that, “Monetary Policy needs a single harmonised index which all over the world is based on consumer prices. The WPI we have is really a producer price index. We may have different indices but headline inflation [in other countries] is always tracked by the consumer (http://www. The finance minister told the Lok Sabha on March 14, 2008 that a new price index is being prepared to measure inflation more accurately. He was referring to a new Consumer Price Index (CPI) for urban India (CPI-U) that is to replace soon the current one for urban non-manual employees (CPI-UNME).1 An internal technical group of the RBI on seasonal movements in inflation recently submitted its report [RBI 2007]. Periodic revision of the weighting diagram and seasonal adjustment are common practices all over the world. But there are other serious problems with our price indices that need attention.Section 1 describes how our five price indices are compiled and their many deficiencies. The deficiencies include importantly the fact that our compilation procedures neither systematically incor-porate new goods and services as they enter the market nor adjust for changes in the quality of goods and services over time. With the opening of the economy to competition from imports and foreign investment and domestic liberalisation, the range of goods and services available to consumers and producers has expanded and their quality has vastly improved. Our price indices, which do not adequately reflect this expansion and quality improvement, could overstate inflation rates and understate the growth in productivity. In the Boskin Commission report (Social Security Administration 1996) and a recent retrospective on it [Gordon 2006], these effects have been shown to be important even in the United States, where procedures for compilation of price indices systemati-cally incorporate new products and adjust for quality improve-ments, and have apparently proved inadequate. Section 2 discusses this report. Section 3 (and also the Appendix) is devoted to a critical evaluation of RBI (2007). Section 4 summarises and concludes with five recommendations inspired by the Boskin Commission report.21 Price Indices and Their DeficienciesFive national series of official price indices are being published in India of which one is the WPI (base 1993-94), and the other four areCPIs consisting of CPI-IW (base 2001) for industrial workers, CPI-AL (base 1986-87) for agricultural labourers,CPI-RL
SPECIAL ARTICLEJune 28, 2008 EPW Economic & Political Weekly218(base 1986-87) for rural labourers, and CPI-UNME (base 2001) for urban non-manual employees. All are fixed weight Laspeyres indices. They play several roles including in the calculation of dearness allowances for government employees, serving as infla-tors (or deflators) for many components of National Income Accounts and importantly for the updating of official poverty lines and hence, for determining levels of poverty and trends at the all India and state levels.3 TheWPI is compiled by the office of economic adviser in the ministry of industry. TheCPI-IW, CPI-AL andCPI-RL are released by the Labour Bureau in the ministry of industry andCPI-UNME is released by the Central Statistical Organisation (CSO). In addition, different states and union territories collect price data and publish price indices. Chapter 11 of the report of the National Statistical Commission [NSC 2001] discusses our price statistics in detail. It noted that the indices compiled by the various states lack uniformity in their base year, item basket, periodicity, segments of population covered, method of compilation and sources of data, and concluded that “it is not possible to under-take proper analysis of regional or state level price variations based on such indices” (para 11.3.2). I will not go into the state level price indices except to note that the problems of the national price indices apply to state level indices also.1.1 Intended Coverage of the CPIsNSC (2001) noted that the different CPIs are meant to be relevant to different segments of the population. CPI-UNME is meant for those who derive 50 per cent or more of their income from gainful employment on non-manual work in the urban non-agricultural sector. The weighting diagram for the index with base 1986-87 was derived from a family living survey conducted during 1982-83 in 59 selected urban centres, with centres being allocated to states in proportion to their 1981 urban population.CPI-RL is for households deriving their major income in the previous year from wage-paid manual employment rural areas. From among rural labour households, those deriving 50 per cent or more income in the previous year from wage-paid manual labour in agriculture are defined to be agricultural labour (AL) households for whom theCPI-AL is intended. Thus, wage-salary-paid non-manual employees and self-employed in rural areas are excluded from these two indices.CPI-IW is for industrial workers employed in any one of seven sectors: factories, mines, plantations, railways, public motor transport undertakings, all electricity generators and distribu-tion as well as ports and docks. The weighting diagram for the index with base 2001 was derived from a working class family income and expenditure survey (FIES) in 78 selected centres in 1999-2000. There is no simple way of estimating the share of population for which each of the fourCPI is meant. A rough upper bound for rural labour is the sum of wage and salary workers and casual labourers in rural areas. Data from the 62nd round of the national sample survey [NSS 2008] show that these two groups accounted for 41 per cent of those usually employed (principal and subsidi-ary status) in 2005-06. For agricultural labour, a rough proxy could be landless agricultural labour and this group accounted for less than 10 per cent of rural households. Reasonable proxies for the categories of industrial workers and urban non-manual employees are not easy to find. If one identifies them very loosely with regular wage and salary workers in urban areas, they accounted for 42 per cent of those who were usually employed. The basic point is that the largest proportion among those who are usually employed, namely the category of self-employed, consisting in 2005-06 of 57 (62) per cent of males (females) in rural areas and 42 (40) per cent of males (females) in urban areas [NSS 2008] are not in any of the population segments covered by the fourCPI. Inclusive of dependents, a proportion of rural and urban population excluded from the ambit of price indices is very large.1.2 Weighting Diagrams and Price DataSince 1950 the NSS has been conducting household expenditure surveys (HES) in the entire country. Yet the weighting diagrams of only CPI-AL andCPI-RL are based on NSS data, with those for CPI-IW andCPI-UNME being based on ad hoc family expenditures in selected urban centres only. The fact that, historically, the ad hoc surveys were used whenCPI-IW andCPI-UNME were initiated, rather than any analysis of the possible use ofNSS data, perhaps explains their continued use. It is worth considering whether to discontinue the ad hoc surveys and instead use NSS data. Also, given that the weighting diagrams for CPI-AL andCPI-RL differ very little from each other and both are based on the same set of retail prices, there is no particular reason for compiling CPI-AL since rural labourers category in any case includes rural agricul-tural labourers as well.CPI-AL andCPI-RL use the same data of monthly retail prices collected from 600 villages spread over 20 states by the field operations division (FOD) of NSS. The NSS has made a proposal to replace the set of 600 base villages by a new set and then increase the set to 1,000 villages, with one-fifth of the sample village being replaced every year. This proposal is yet to be implemented. For CPI-UNME monthly retail prices are collected byFOD in respect of selected goods and services from related markets in 59 centres. For CPI-IW retail prices are collected by the Labour Bureau through part-time price collectors on weekly, monthly and half- yearly basis and from selected markets and shops. It is unclear how one would define what should constitute the universe of wholesale or retail prices from which a “representative” set is to be chosen for use in compiling price indices, let alone analyse whether or not our procedures of collection would yield a “repre-sentative” set. Although some rules of thumb are specified for the replacement of outdated/unavailable centres as well as shops and commodity specifications, in practice there is a considerable lag in replacements. Furthermore, the prices actually paid by buyers often differ from, and are higher than, the maximum retail prices (inclusive of local taxes) that the shopkeepers are supposed to charge, although the latter naturally report to the data collectors only the prices they are supposed to charge. TheWPI is compiled on a weekly basis based on the price quota-tions collected by official and non-official agencies on 435 selected items identified in the basket of the index. The NSC (2001) had this to say on the prices that go into the index: “Owing to the
SPECIAL ARTICLEEconomic & Political Weekly EPW June 28, 2008219wide variety of sources, centres and specifications and due to the practical compulsion of collecting data by the voluntary method, it is difficult to maintain uniformity in the concept of wholesale price in the collection of price data. In many cases, these prices correspond to farm-gate, factory-gate or mine-head prices; and in many other cases they refer to prices at the level of primary markets, secondary markets or other wholesale or retail markets theWPI as presently complied does not reflect either the producer price or market price in a consistent manner” [NSC 2001, para 11.4.1]. Besides, the weighting diagram of the WPI apparently reflects domestic absorption of goods (domestic production plus imports minus exports). For both reasons contrary to the belief of the governor of the RBI, WPI is not strictly a producer price index. The voluntary method allows the options of reporting online, by email or fax. Yet compliance and quality of reporting are far from satisfactory. The problems are compounded by understaffing of statistical agencies.1.3 Service Sector in Price Indices4TheNSC commented critically on the many problems with pricing services. Currently services are not included in the WPI. Some of the pricing problems are inherent to services, such as those arising from variations in quality that are often difficult to observe let alone quantify. With the service sector (excluding construction) now accounting for a growing share of GDP (around 55 per cent in 2007-08), and more and more services being traded domestically and internationally, including services inWPI is essential. Compared to its share of 55 per cent inGDP, the weight of services inCPI-UNME is 24 per cent, inCPI-IW 16 per cent, and 12 per cent in bothCPI-RL andCPI-AL. It is unlikely that the cost of the services included in various CPI (for example, amusement, personal care, medical services, education, etc) can be or are well estimated. The soon to be released new monthly producer price index will include services. A technical group is currently grappling with the problems of pricing services.5 Following recommendations of UNSNA (1993), Indian national accounts statistics also use the concept of “Finan-cial Intermediation Services Indirectly Measured” or FISM. Some of the issues relating to estimation of FISM of the financial sector as a whole and its allocation across components of the sector are discussed in the papers of Chakrabarty and Das (2007) and of Barman and Samanta (2007). For brevity I will not go into them here. 2 Boskin Commission ReportThe finance committee of the US Senate appointed in June 1995 an advisory committee to study the CPI, chaired by Michael Boskin of Stanford University. The Boskin Commission (BC) submitted its report (Social Security Administration 1996) on December 4, 1996. Since its publication, some of the recommen-dations have already been implemented. The major finding of the BC was that theCPI overestimated the underlying inflation by as much as 1.1 per cent per year, with a plausible range of 0.80-1.60.6Most of the bias (0.60 per cent) came from inadequate account-ing for new products and quality change in existing products. This finding is very relevant in the Indian context for two main reasons. First, unlike in theUS, the procedures of compilation of our price indices do not systematically or otherwise allow for the introduction of new products or quality changes, except insofar as might come about incidentally when the base-period of an index is updated. Second, prior to reforms of the 1980s and 1990s, prices of several commodities were controlled or administered and the licence-permit raj severely restricted the range of products available in the market to consumers and producers through import licensing, tariffs and quotas, capacity licensing, Monopolies and Restrictive Trade Practices (MRTP) regulations and small-scale industry reservations. The reforms have either abolished or considerably relaxed most of these restrictions. One of the, if not the, most important consequence of reforms is that competition has vastly increased in the economy. Not only the range of products available in the market has increased with improvement in the quality but competition has led to reductions in their real costs (adjusted for quality). For all these reasons, it is likely that our CPIs and WPI, which do not reflect the changes, considerably overstate the underlying inflation since the reforms. There are procedures that are in place in the US for taking into account introduction of new products and for quality change in the existing products – in fact the BC found these procedures wanting. Unfortunately, we do not seem to have any such system-atic procedures.OurCPIs (and WPI) like the USCPI, are fixed weight Laspeyres indices and they do not account for the substitution away from commodities whose relative prices are increasing and towards those with falling relative prices. Inflation as measured by Laspeyres index would, therefore, be an upper bound for the underlying inflation that took into account consumer substitu-tion. The BC’s estimate of this overestimation is 0.40 per cent per annum. The failure to account for lower level substitution among different varieties of the same product (e g, different varieties of bananas) is 0.25 per cent. The failure to account for higher level substitution across different products within a broad category (e g, between bananas, guavas, oranges, etc, among fruits) is 0.15 per cent.7 Although similar substitution issues arise with ourCPIs as well, their quantitative significance need not be the same. I return below to the empirical evidence on this based on Deaton (2005, 2008).TheBC distinguished the concept of a cost of living index (COLI) from a CPI and recommended that establishing a COLI should be the objective in measuring consumer prices. A COLI compares the minimum expenditure required to achieve the same level of welfare across two different sets of prices, given a utility function that summarises consumer preferences. The COLI could in principle be defined for each individual or household based on its preferences and the prices faced by it. This, however, is obviously impractical. Instead, it would be defined for a repre-sentative household. TheCPI on the other hand, prices a fixed but representative basket of goods and services over time. For the CPI andCOLI to be identical, given the same basket of goods and prices in both, expenditure shares have to be independent of price changes. Only if the underlying utility function of the representative consumer is of the Cobb-Douglas form, will this independence hold.
SPECIAL ARTICLEJune 28, 2008 EPW Economic & Political Weekly220TheBC also identified another source of upward bias in the CPI based inflation rates, called outlet substitution bias, arising from the failure of CPI procedures to take into account fully a shift in purchases to lower price outlets, following the transformation of retailing, with superstores, discount stores and the like accounting for a growing fraction of sales relative to a decade earlier. Indian retailing is yet to be transformed. Still, the fact is that large corporations have entered the retail sector in the last decade or so and more, including global chains like Walmart with its Indian partner are poised to enter suggests that a transformation is taking place, particularly in urban areas. It is also the case that both prices as well as qualities of products sold in supermarkets and other corporate retail outlets differ from those of comparable products in small-scale neighbour-hood stores in India. Thus,theoutletsubstitutionbias will be compounded by our failure to make adjustments for quality changes. Two recent papers of Deaton (2005, 2008) illuminate the implications of the use of different price indices. Deaton (2005) uses the consumption data from the 43rd, 50th and 55th rounds of NSS to compute a range of price indices for 1997-2000 relative to 1993-94 and for 1993-94 relative to 1987-88, for rural and urban sectors of each of the large states and for India as a whole. The price indices were derived using the expenditure share on each good by each sample household averaged by sector and state. The unit value of each good was obtained by dividing the expenditure on it by the quantity consumed, taking care not to include categories with heterogeneous items.8 The median unit valuetoeachstate and sector was defined as the price of that good. Thus based on the vectors of expenditure shares in each of the rounds, four price indices (Laspeyres, Paasche, Fisher Ideal and Tornqvist) were computed. The deflator used in updating the official poverty line – the PLD – is compared with four indices derived from NSS data. For 1999-2000, compared to 1993-94, the rural Laspeyres index fromNSS showed less inflation than thePLD at the all India level and the other three indices from NSS showed even lower inflation than thePLD deflator. In contrast, in the urban sector, the Laspeyres Index is slightly larger than but close to the PLD. For 1993-94 as compared to 1987-88 also the Laspeyres Index showed a similar pattern. However, Deaton’surban price levels (not to be confused with inflation rates) are consistently around 15 per cent higher than the rural levels in all three rounds. On the other hand, in the official PLD, the urban price levels were higher by around 40 per cent than rural price levels. Both the rural-ur-ban differences in price levels between Deaton’s price indices and the PLD, and whether or not one adjusts for the comparability problem with the 55th round (1999-2000), affect the comparison of poverty estimates. Deaton argues that the urban official PLD are implausibly high. Deaton’s poverty estimates using Tornqvist price indices and adjusting for the overstatement of expenditures in the 55th round and official estimates are shown in the table.As expected, the choice of unit values as prices (rather than retail prices as in the official PLD) and the price indices used by Deaton have a significant impact on poverty estimates.Deaton (2008) compares the 61st round (2004-05) with the 55th round (1999-2000) using NSS survey unit values and expend-iture shares for putting together a food price index and combin-ing it with the non-food price index from CPI-AL for rural areas andCPI-IW for urban areas to derive an all-commodities price index. The survey-based food price index and food components of CPI-AL (for rural) andCPI-IW (for urban areas) closely match each other from the 38th round (1983) to the 55th round (1999-2000) and diverge thereafter. The survey-based rural food price index with the 55th round rose to 111.21 by 2004-05 while the food component CPI-AL rose only to 106.67. In urban areas, the survey-based index rose to 113.90 while the food component rose to 112.88. Thus food component of CPI-AL understated inflation between 1999-2000 and 2004-05. Taking into account this understatement of food price inflation inCPI-AL and its overstatement of the expenditures share raises the rural poverty head count ratio to 30 per cent, as compared to the official estimate of 27.3 per cent. The two papers of Deaton clearly demonstrate the sensitivity of poverty estimates of the choice of price deflators. 3 Inflation Rates and Seasonal AdjustmentThe internal technical group appointed by the RBI had the follow-ing terms of reference (TOR): (i) To suggest appropriate methods to estimate seasonal factor to be used to gauge the inflationary pressure in the Indian economy during the year.(ii) To analyse seasonal behaviour of inflation in India cover-ing the CPI andWPI, including at the disaggregated level.(iii) Any other issue as deemed necessary for the analysis of seasonal behaviour of inflation.Unfortunately, the report concentrates essentially on the arith-metical and purely statistical aspects of price indices, inflation rates and seasonal adjustments. It does not discuss any theory of either the determination of the path of the price index (or equiva-lently the associated inflation rates) or the appropriate price indices relevant for consumers or producers and on which they presumably form expectations. In its executive summary, the report states: “Communication of inflation rate appropriately is important, as economic agents and financial markets are expected to form their price expectations based on official published figures”. But nowhere in the report is there any discussion in the Indian context of what prices are relevant for different agents and about which they form expecta-tions (static? adaptive? rational?) and over what time horizon. Even more disappointingly for a report of an internal committee of the RBI, the problem of the central bank communicating its policy stance relating to inflation is not even mentioned. In the entire report there is no reference to the economics of inflation. To be fair, the TOR did not explicitly require the committee to go Table: Headcount Ratios of Poverty 1987-881993-941999-2000 Rural Urban Rural Urban Rural UrbanOfficial 39.4 39.1 23.5Deaton 39.0 22.8 32.918.125.3 12.5Source: Deaton (2005), Tables 17.5 and 17.6.
SPECIAL ARTICLEEconomic & Political Weekly EPW June 28, 2008221into the economics of inflation including theories of expectation formation. On the other hand it was not precluded from doing so, had the committee “deemed necessary for the analysis of seasonal behaviour of inflation”. Unfortunately, it did not.On seasonality, volatility, etc, there is nothing methodologi-cally new in the report. In some countries, “core inflation” is distinguished from “headline inflation” by excluding some commodities (e g, food, oil) whose prices are known to be partic-ularly volatile in defining the former. In the Indian economy, the report asserts that “all the price series in India (viz,WPI, CPI) have peak/trough months around the same time of the year, mainly due to seasonality in food prices, which are the major source of seasonality in all price indices” (para 7.7). If this is indeed the case one approach to seasonal adjustment is not only to exclude food prices, but also the component of other prices that are predictable given food prices. The report speaks only of the exclu-sion of food prices.The report discusses the susceptibility of annual point-to-point inflation rates to what it calls the “base effect”. It suggests that an “alternative method of measuring inflation could be month-over-month variation inWPI, using seasonally adjusted data” and claims that such an inflation rate “exhibits high volatility and does not exhibit appropriate stable signals about price changes” (Executive Summary). Since the report is silent on what would be the “appropriate stable signals” for different agents, it is not possible to evaluate this claim. The report somewhat exaggerates the seriousness of both the “base effect” and of the volatility in month-over-month variation in seasonally adjustedWPI. It is far from obvious that they are particularly relevant purely from the perspective of “communi-cating” inflation rates appropriately. Although the report does not define precisely what it means by the “base effect”, its intended meaning is clear from its discussion of it. The “base effect” is a purely statistical effect that makes annual point-to-point inflation rates to exhibit wide and abrupt variability, at times within the year, even when the movements in the corre-sponding price index do not exhibit such variability. Thus, in Section II.2.1 it is pointed out that “while the price index was stable during [the months of] the year [2001-02], there was wide variability in the annual point-to-point inflations due to its volatil-ity in the previous year (base effect). A similar phenomenon was observed recently. From September 2006 to January 2007, the annual point-to-point inflation increased from 5.41 per cent to 6.69 per cent, and as a result, a public opinion was formed that the inflation in India was becoming too high, and perhaps was moving beyond the control (of monetary authority). However, during the same period,WPI increased very moderately from 208.3 to 209.0” (para 2.8). No evidence is cited, however, on the formation of public opinion or of the perception that inflation was beyond the control of the monetary authority.The following hypothetical example, which is in fact a stylised version of ChartsII.1 andII.2 in the report, is helpful in under-standing the base effect. Suppose, say in year 2, the WPI was constant, i e, it had the same value in each week from the first week of January to the last week of December and as such, the week-to-week(WtW) inflation rate was zero in each week from January to December. If only for simplicity, we assume that the WtW value ofWPI in the first week of January in year 2 was the same as its value in the last week of December of year 1, theWtW inflation rate in the first week of January of year 2 would also be zero. Suppose further, in year 1, the WPI rose each week, then it is clear that the annual point-to-point inflation rate willfall in every week of year 2!This example illustrates the general point that the movement in the annual point-to-point inflation rates as the reference week moves within a year is the result of not only the pattern of movement of the WPI from week-to-week in that year, but also the pattern of movement from week-to-week in the preceding year as well. Thus focusing on movement of annual point-to-point inflation rates across weeks of a given year combines or confounds, to put it simply though not precisely, within year pattern of movements inWPI each year and the shift in this pattern between consecutive years. This problem could arise also if one were to use seasonally adjusted monthly WPI rather than unadjustedWPI. However, if seasonal adjustment was purely proportional, i e, seasonally adjustedWPI for weekw in any yeart is a (w) times the unadjusted WPI for the same week, then clearly the weekly point-to-point inflation rates would be the same whether one uses seasonally adjustedWPI or not. The report has not recognised that there is an arithmetical relationship between the annual point-to-point annual inflation rate in any given week and the wtw inflation rates in the 52 weeks preceding that week:9 the annual point-to-point inflation factor (1 + inflation rate) at any week of a year is a geometric average of the week-to-week inflation factors during the previous 52 weeks. Thus, the annual point-to-point rate in any week is a summary measure of the path of the week-to-week inflation rates in the previous 52 weeks prior to that week. This means that as one varies the week of reference for the annual point-to-point infla-tion rate, the rate could change in an arbitrary fashion. Thus as the reference week is moved from the first week of April to the last week of March in a fiscal year, the point-to-point annual inflation rate would vary arbitrarily if the time-paths of the week- to-week inflation rate in the 52 weeks prior to the reference week vary in an arbitrary fashion. This is an elementary arithmetical fact that in my view need not have any impact on communication of inflation rates or formation of price expectations as long as it is understood by all agents, as such.If the weekly seasonally adjusted price index is proportional to the unadjusted price index and the constant proportionality for any given week is the same for all years, then the point-to-point annual inflation rates would be the same whether or not a season-ally adjusted price index is used. If we measure volatility by log variance, the volatility in week-to-week inflation factors within a year again does not depend on whether a seasonally adjusted series is used. However the volatility in the point-to-point annual inflation factors within a year would indeed depend on whether or not seasonally adjusted price index is used. It is also the case that in general seasonal adjustment could change volatility in either direction, up or down.
SPECIAL ARTICLEJune 28, 2008 EPW Economic & Political Weekly222TheRBI report claims that “it is useful to note that the consum-ers are affected by the actual price level (i e,WPI) and not inflation rate. Similarly, for converting many macro variables from their nominal to the real values, price levels are used” (para 2.9). Apparently the authors have forgotten that the price level indicated by a price index such as the WPI is relative to the base period at which the price level is set at 1 (or 100). Therefore,WPI at any time point could equivalently be described as the inflation from the base period to that time point in it. Put another way, the base year basket of commodities is the numeraire. The problem of conversion of nominal to real values through deflation by a price index is not about an essentially arbitrary choice of a base period basket or numeraire, but about the choice of the appropri-ate price index that ensures real income changes track changes in consumer welfare. Long ago Samuelson and Swamy (1974) addressed this issue. Diewert and Nakamura (1993), Chapters 8 and 9 provide a thorough analysis of the issues.Equally inexplicable is para 2.10 of the report: “While there can be several reasons for abrupt changes in domestic inflation rate, the major sources of its volatility include sudden domestic supply shock (e g, prices of agricultural commodities), inter-national demand/supply conditions (e g, global crude oil/steel/foodgrains prices) and base effect. While the first two shocks affect the price level, the third affects only the inflation rate in which case the volatility introduced by it should not be alarming. In case such incidents are not explained properly, there is a risk that economic agents may form an opinion of (actual) price rise in the economy.” This statement again is confused about price levels and inflation rates. After all, if supply shocks affect the price level (i e, WPI), at some point of time, then they have to affect the infla-tion rates at that point of time relative to the past, as well!. 4 SomeProposalsforReformAny proposal for reform of CPIs, to be feasible to implement, has to address the following features of the Indian economic scene. First of all, consumption from home produced commodities is significant for a large segment of rural households, namely culti-vator households. Second, consumption of foodgrains of some cultivator households, particularly from the segment of small and marginal farmers, is largely from home-grown stock during the peak harvest season, but in the lean season the same households purchase their consumption from the local market. This means that for such households the relevant prices for computing their cost of living is farm harvest prices representing the opportunity cost of their consumption from home-grown stock during peak harvest season and retail market prices during the lean season. However, the prices that go into the compilation of the two rural CPIs, namelyCPI-RL andCPI-A, are the local retail market prices collected from 600 villages. While consumption out of home-grown stock is unlikely to be significant for AL it is not necessarily the case for RL since some cultivator households might be engaged in off-farm manual labour in off seasons. Third, it is unclear what in principle the universe is of commodities whose wholesale or retail prices are being collected, given that over time some commodities disappear from the market altogether, others change enough in other dimensions including quality to be deemed different from what they used to be before, and new commodities that did not exist before enter the markets. There is no way of evaluating how “representative” the prices are that go into our price indices because of the changing universe of commodities. Even if it did not change, the procedures of price collection are by no means through a well-defined probability based sample design. I propose five reforms. First is a proposal for a study of the likely quantitative significance of these issues as well as research for designing a procedure for accommodating changes in quality, the structure of outlets, and the entry into the market of new commodities/varieties and exit of others.Second, in the spirit of Boskin Commission’s ideas, I propose, first, that the objective of establishing a cost of living index (COLI) should be the goal of measuring consumer prices. AlthoughBC rejected devising more than one COLI for different subgroups in the US, in our context of wide diversity in socio-economic-cultural-regional characteristics of our population, we have to consider several COLI, each appropriate to a particular sub-group of the population. Third, I propose replacing WPI with a producer price index and modifying it to reflect systematically changes in quality, new commodities and also including services in it. The weighting diagram will be based on the production pattern. It remains to be seen how far the new producer price index will meet these requirements. The sources of price data that go into the index have to be re-examined to make sure that prices reflect the concept of being wholesale prices and also are reasonably repre-sentative. To do all this satisfactorily, will have to await the results of the study of my first proposal. But in the meantime ad hoc changes can be explored. Also, changing the base year every five years or so would help.Fourth, I propose that instead of fourCPIs, two (urban and rural) retail price indices for India as a whole, and also for the states and union territories should be compiled. The intention is that the retail price indices would be the purchasers’ counterpart of producer price index with a weighting diagram that reflects the patterns of retail expenditures, rather than the pattern of production as a producer price index would. The details of the appropriate weighting diagram and the sources of price data have to be carefully thought through. Their frequency would be the same as that for our current CPIs. Given their high frequency, they could be used for measuring retail-price inflation.Fifth, I propose fourCOLI one for each of two sub-groups of population in rural and urban areas, to be put together based on data from “thin” and “thick” rounds of NSS. Naturally they would be published annually with a lag of at the most two years. Being COLI instead of retail price indices, these have to approximate the true cost of living for the relevant sub-groups of the population. At first blush, it might seem natural to think of four sub-groups: poor and non-poor in rural and urban areas respectively. However, this is not easy to do since for defining the relevant poverty lines one would need a price index in the first place. The alternative is to devise a classificatory scheme that uses charac-teristics that are independent of prices, income and economic status. One such scheme is to divide rural and urban populations
SPECIAL ARTICLEEconomic & Political Weekly EPW June 28, 2008223into two groups, one including SSC/ST and backward castes and the second including all others. Although this division does not correspond exactly to the conventional below the poverty line (BPL) and above the poverty line (APL) divisions, it is likely to be close enough. I understand that the HES schedule of theNSS has information on the caste (other thanSC/ST) status of the house-holds. Other ways of approximatingBPL/APL groups should also be explored. Assuming that four groups can be identified, usingHES data of NSS, by adapting Deaton (2005) methodology one could put together price indices for each group every year at the “All India” level using the quinquennial (thick) and annual (thin) rounds and at the state/UT level every five years using quinquennial rounds. Because the Deaton methodology uses the expenditure patterns and unit values from theHES, the price indices are closer approximations to the “true”COLI. In proposing this, I am not ignoring the well known problems with the use of unit values as prices since they incorporate quality and price effects, and the quality effects are likely to be correlated with expenditure levels. I am also aware that many non-food commodities and services are very heterogeneous. For this and other reasons HES collects only expenditure data but not quantity data on many non-food items and usable unit values for them cannot be derived fromHES. Perhaps a way could be found of usingretailpricesfor them with expenditure shares fromHES. The advantage of the unit values, such as they are, is that they come closer to address-ing quality and new product effects and also the varyingcosts and shares of home produced goods in the consumption basket. I suggest that some illustrative exercises be done using recent rounds of NSS data. Hopefully the new Index for urban areas proposed to be introduced by theCSO to replace CPI-UNME presumably will be usingthe Deaton (2005) methodology or some version of it.Notes 1 The WPI is also to be replaced gradually with new indices on service prices and prices at the produc-ers’ level. These indices would be released every month rather than every week as the WPI, with price movements of essential commodities would continue to be released weekly. Also, from May 2008, actual price quotations would be released monthly. The new producer price index will comprise 980 items as against 435 in the current WPI and will have 2004-05 as bas year. The new index will include new items and drop some old ones These are indeed positive developments (http://www. hindu. com/2008/05/01/stories/ 2008050155681700.htm ) 2 This article draws on Srinivasan (2008a). 3 The Sixth Pay Commission has apparently recom-mended the compilation of a new price index for government employees. I thank Suresh Tendulkar for drawing my attention to this fact. 4 There are some serious issues of measurement of value added by the service sector in our national accounts. A discussion of some of them could be found in the papers published in the special issue on services in theEconomic & Political Weekly, September 15, 2007. I discuss others in Srini-vasan (2008b). 5 I thank Suresh Tendulkar for this information.6 Robert Gordon (2006), who was a member of the commission, in his retrospective on the BC report and its critiques, argued that, first, BC’s bias estimate of 1.1 per cent per year should have been 1.2-1.3 per cent, with BC overestimating the bias due to quality change and new products by 0.2 per cent which is more than offset by its underestimation of upper level substitution bias by 0.3-0.4 per cent. Second, Gordon estimates that the upward bias in CPI has declined from its corrected value of 1.2-1.3 per cent in 1995-96 to around 0.8 per cent around 2005-06. 7 The theory of index numbers, including substitu-tion issues and desiderata (e g, time reversal test) that index numbers ought to satisfy are well known. See Erwin Diewert and Alice Nakamura (1993).8 In the 55th round, for example, quantity and expenditure data were collected for 173 items (mostly food, fuels and intoxicants). Still the fact that categories with heterogeneous items (partic-ularly non-food categories, such as services, durables, etc) account for significant shares of expenditures have to be kept in mind. Deaton (2008) combines unit values for food items from the NSS survey with the index of prices of non-food items from CPI-AL or CPI-IW.9 The appendix in Srinivasan (2008b) establishes this result and the one in the next paragraph on volatility of inflation rates. ReferencesBarman, R B and G P Samanta (2007): ‘Measuring Banking Intermediation Services: Issues and Challenges for India’ in EPW, “Special Themes”, 42(37), September 15, pp 3754-62.Chakrabarty, A B and A Das (2007): ‘Banking Sector’s Output in National Accounts: Measurement Issues’ in EPW, “Special Themes”, 42(37), Septem-ber 15, pp 3764-68.Deaton, A (2005): ‘Prices and Poverty in India, 1987-2000’, Chapter 17 in Angus Deaton and Valerie Kozel (eds),The Great Indian Poverty Debate, MacMillan University Press, New Delhi. – (2008): ‘Price Trends in India and Their Implica-tions for Measuring Poverty’, Economic & Political Weekly, 43(6), pp 43-49.Diewert, W E and A O Nakamura (eds) (1993): Essays in Index Number Theory, Vol I, North Holland, Amsterdam.Gordon, R (2006): ‘The Boskin Commission Report: A Retrospective One Decade Later’, Working Paper 12311, National Bureau of Economics Research, Cambridge, MA.NSC (2001): Report of the National Statistical Commis-sion, Government of India, Ministry of Statistics and Programme Implementation, New Delhi.NSS (2001): Employment and Unemployment Situation in India, 1999-2000, Part II, Report # 458, National Sample Survey Organisation. – (2006): Employment and Unemployment Situation in India, 2004-05, Part I, Report # 515, National Sample Survey Organisation. – (2008): Employment and Unemployment Situation in India, 2005-06, Report # 523, National Sample Survey Organisation.RBI (2007): Report of the Internal Technical Group on Seasonal Movements in Inflation, Reserve Bank of India, Mumbai.Samuelson, P and S Swamy (1974): ‘Invariant Economic Index Numbers and Canonical Duality: Survey and Synthesis’, The American Economic Review, 64, pp 566-93.Social Security Administration (1996):Toward a More Accurate Measure of the Cost of Living, Final Report to the Senate Finance Committee from the advisory Commission to Study the Consumer Price Index ( Srinivasan, T N (2008a): ‘Some Aspects of Price Indices and Inflation Rates and National Income Statistics’, forthcoming in theV K R V Rao Cente-nary Commemorative Volume, Institute for Social and Economic Change, Bangalore.–(2008b): ‘Services Sector in National Accounts Statistics’, Economic Growth Centre, Yale University.UNSNA (1993): System of National Accounts, United Nations, New Yor’k.SPECIAL ISSUEReview of LabourMay 31, 2008Class in Industrial Disputes: Case Studies from Bangalore –Supriya RoyChowdhuryEmployee Voice and Collective Formation in Indian ITES-BPO Industry Philip Taylor, Ernesto Noronha, Dora Scholarios, Premilla D’Cruz The Growth Miracle, Institutional Reforms and Employment in China –Ajit K GhoseSoccer Ball Production for Nike in Pakistan –Karin Astrid SiegmannLabour Regulation and Employment Protection in Europe: Some Reflections for Developing Countries –A V JoseLabour, Class and Economy: Rethinking Trade Union Struggle –Anjan Chakrabarti, Anup Kumar DharFor copies write to: Circulation Manager Economic and Political Weekly,320-321, A to Z Industrial Estate, Ganpatrao Kadam Marg, Lower Parel, Mumbai 400

Dear Reader,

To continue reading, become a subscriber.

Explore our attractive subscription offers.

Click here

Back to Top