The recent divergence of different measures of inflation is an important issue that has been discussed in literature over the last couple of years. Conceptually, wholesale price index (WPI) and consumer price index (CPI) are different as they capture different sets of baskets of goods and services as well as different stages of the supply chain.
The measure of inflation is important for policymakers, especially the central banks that have formally adopted an inflation targeting framework. The Reserve Bank of India (RBI) adopted a similar framework in 2016 and recently announced a review of the performance of the same. This review comes at a time when the monetary policy committee (MPC) has been criticized for acting slow on repo rate cuts while real interest rates in the country remain amongst the highest in the world, despite moderation in inflation well below the target. Bhalla and Bhasin (2019)1 look at this issue in detail as they argue the need to revisit our current framework.
However, an inflation targeting framework is only as good as the measure of inflation. That is, an inflation indicator that systematically overestimates inflation will result in a tighter monetary policy than needed while systematic underestimation will result in significant monetary accommodation. Such systemic problems can have unintended consequences and therefore, it is important to have a detailed discussion on whether our current inflation indicator adequately captures inflation.
This issue is important and at the heart of the debate is the choice of indicator for inflation targeting with former Chief Economic Advisor Dr. Arvind Subramanian preferring the wholesale price index while former RBI Governor Dr. Raghuram Rajan advocating for using consumer price index. The rationale for WPI was that it attributed a lower weight to food than CPI which meant that it was more suitable for the conduct of monetary policy which has little impact on food inflation. Dr. Raghuram Rajan, however pointed how CPI is more important and inflation expectations are based on CPI rather than WPI making it the preferred indicator for inflation targeting. Indeed, most countries with inflation targeting framework use CPI as the inflation target. However, it is important to recognize that the high share of food can lead to a situation where the overall CPI may be higher because of an increase in food inflation. We witnessed this from November 2019 to February 2020.2
As the RBI reviews the Monetary Policy Framework, we must also review our indicator of inflation and address some critical issues associated with them. The structure of the paper is as follows. In section 1 we discuss the procedure to arrive at CPI estimates and the implications of the non-release of the 2017 Consumption Expenditure Survey. This is followed by a section where we present an alternative CPI series and discuss the differences between this series and the official series. A detailed discussion is needed on how to construct an appropriate inflation indicator that can be used for the purpose of our inflation targeting framework as several authors have argued for a multi-indicator approach. This is followed by a conclusion.
The recent consumption expenditure survey (CES) (2017) has been withheld by the government as it is inconsistent with administrative data. There is adequate evidence that illustrates that the findings of CES (2017) are indeed inconsistent and that the government is right in withholding the release of the report. Bhalla and Bhasin (2019)3 and Bhalla, Bhasin and Virmani (2020)4 discuss some of the issues with the Consumption Expenditure Survey in detail.
While the findings are inconsistent, the government must release the unit level data for academics to see the problems associated with the dataset. A related issue is of revising the weights for our consumer price index. In India, consumer price inflation is arrived by using the weights that are obtained from the National Sample Survey Office (NSSO)’s consumption expenditure survey which helps in creation of a representative basket of goods and services. In the absence of the release of the CES data, such revision of weights is not possible. Therefore, we may have to wait for 2021 or even beyond to readjust the weights of our CPI index as such an adjustment is only possible after the next round of such a survey.
The issue that is also important is of the inability of our CES to capture service sector consumption and this has been mentioned by the Adhikari Committee (2015) report.5 Indeed, the NSSO’s CES survey results often underestimate the expenditure on social consumption such as health and education compared to NSSO’s focussed surveys on social consumption. It is well regarded that there’s some degree of underestimation of consumption in these surveys and this mostly corresponds to expenditure on services.
This is precisely why a discussion on appropriate weights for CPI is important as over time as economies develop, the share of consumption on food goes down (same is true for goods) while the share of consumption of services starts to increase. Therefore, our inability to capture the expenditure on services can have implications for the weights of CPI series. There has been a committee that has been constituted to address some of these issues and it will submit its report shortly, but in the meantime, we must address the issue of appropriate weights for our inflation indicator.
The other alternative is to use the National Account Statistics (NAS) which gives us an item-wise estimate of consumption expenditure on an annual basis. The weights between 2011 and 2017 have changed and we represent this in Table No 1. It is worth noting that the share of food has declined by 4 percentage points while the share of others has gone up by around 4 per cent points while of miscellaneous it has gone up by 7 per cent points. This shows that indeed the consumption basked in 2017 differs from the one in 2011 and therefore, the weights for CPI in 2017 must reflect this new basket.
This makes it obvious that an increase in food prices can have a bigger impact on the inflation figures while in reality, the cost of the representative consumption basket may witness lower inflation. It is possible to use these weights to arrive at inflation figures for various months and to contrast it with the official series. Given that we do not have a consumption expenditure survey to rebase our CPI in 2017, it may make sense to instead use the national account services weights to arrive at newer weights for the same.
Table 1: Share of Consumption Expenditure (2011 and 2017)
|NAS (2011)||NAS (2017)|
|Cereals and products||5.9||5.2|
|Meat and fish||2.4||2.4|
|Milk and products||6.5||4.7|
|Oils and fats||2.0||1.3|
|Pulses and products||1.5||1.3|
|Sugar and Confectionery||1.5||1.1|
|Prepared meals, snacks, sweets etc.||2.4||2.3|
|Food and beverages||32.7||28.5|
|Pan, tobacco and intoxicants||2.8||1.8|
|Clothing and footwear||6.3||6.4|
|Fuel and Light||4.7||4.1|
|Household goods and services||3.2||3.6|
|Transport and communication||17.4||19.0|
|Recreation and amusement||1.0||0.8|
|Personal care and effects||1.9||1.9|
|Note- NAS refers to national account statistics for the respective years|
Source: Author’s Computation based on data from National Accounts, various years
An alternative indicator of inflation can be obtained by using the share obtained from national account statistics. We extend the methodology used by Bhasin (2020)6 as we use these shares to arrive at appropriate weights and use the monthly CPI data to arrive at alternative estimates. The finding suggests that there have been times where the official CPI is overestimating inflation while at other times, it tends to underestimate inflation.
The findings are presented in figure 1 which has the official CPI, the national accounts (2011) weights-based inflation and the national accounts (2017) weights-based inflation. Based on the results, we rule out the possibility of either systemic over-estimation or systemic under-estimation of inflation.
Figure 1: Inflation based on different weights
Source: Authors Computation
The other important finding is that there is little difference in the 2011 and 2017 weights-based estimator for inflation. However, there is some difference between the official series and the national account statistics-based series, and this difference is primarily driven by the higher weights to food in the official series. Therefore, inflation from April 2018 till January 2019 was higher than the official series. Similarly, for the month of November 2019, December 2019 and January 2020 the official CPI inflation was higher than the national account statistics weights-based inflation. This was driven primarily by high food inflation which had a higher weight in the official series. It is interesting to note that this surge in inflation came at a time when WPI was between 1 and 3 [refer figure 2].
Figure 2: CPI and WPI Inflation
Such divergence in inflation figures suggest a need to revisit the debate regarding the choice of inflation target which is used for our monetary policy framework. While many have argued for a multi-indicator approach, a switch to non-food CPI inflation with the appropriate weights based on the national account statistics could also serve as an efficient indicator for inflation.
In the absence of 2017 CES weights, we will have to wait till 2021 before we rebase our Consumer Price Inflation. However, the consumption basket will shift significantly between 2011 and 2021 which makes it important to address the issue to avoid chances of over or underestimation of inflation. While we find no evidence of either systemic over or underestimation of inflation, however, the recent spike in food inflation has resulted in an acceleration in official CPI well beyond the 2-6 per cent range even as the other inflation indicators have remained broadly within it. A higher official print may also result in an impact on inflationary expectations and therefore may result in judgment errors while conducting monetary policy. Consequently, there is a need to address this issue as a part of the current review of the monetary policy framework that is based on an inflation targeting regime. A wider debate is needed on the need to move towards a CPI inflation that is based on weights derived from the National Accounts Statistics.
 “Opinion | Is the NSO’s Consumption Data for 2017-18 beyond Salvation?” Accessed March 24, 2020. https://www.livemint.com/opinion/online-views/is-the-nso-s-consumption-data-for-2017-18-beyond-salvation-11574962245323.html.
 “Retail Inflation Rises to 7.59% in January on Higher Food Prices.” Accessed March 24, 2020. https://www.businesstoday.in/current/economy-politics/retail-inflation-rises-to-7-59-percent-in-january-on-higher-food-prices/story/396045.html.
 “The Results of the NSO Survey 2017-18 Are Truly Bizarre | The Indian Express.” Accessed March 24, 2020. https://indianexpress.com/article/opinion/columns/nso-unemployment-job-data-statistics-6138396/.
 “Only 8.4 Crore Poor in India, Claims a New Study – The New Indian Express.” Accessed March 24, 2020. https://www.newindianexpress.com/business/2020/feb/18/only-84-crore-poor-in-india-claims-a-new-study—2105127.html.
 “Documents / Report of the Sub- Committee for National Income | Ministry of Statistics and Program Implementation | The Report of Prof. A. K. Adhikari Committee on PFCE. Government Of India.” Accessed March 24, 2020. http://mospi.gov.in/publication/documents-report-sub-committee-national-income-0.
 Bhasin, Karan. “CPI Weights Need Rejig for Gauging Inflation Better.” Livemint, February 20, 2020. https://www.livemint.com/news/india/cpi-weights-need-rejig-for-gauging-inflation-better-11582218025628.html.