Issues in construction: base year, representativeness and base revision

Detail

1. Why construction matters

• An index number measures how a group of related things (prices, or production) changes over time. It compares each period with a reference period.
• A poorly built index gives a wrong reading of inflation or growth. That wrong reading then flows into policy: RBI interest-rate decisions, dearness allowance (DA) for workers, and real GDP.
• NCERT (Class 11, Index Numbers) gives a five-point checklist for building an index: 1. a clear purpose 2. representative items 3. a "normal" base year 4. the right formula 5. reliable data

2. Clear purpose

• First decide what the index should measure.
• Volume index: measures change in physical quantity, e.g. IIP (Index of Industrial Production).
• Price index: measures change in prices, e.g. CPI and WPI.

• Value index: measures change in price × quantity together.

• A volume index is the wrong tool when you need a value index. Always match the tool to the question.

• The IIP is officially "a composite indicator designed to measure the changes in the volume of production" of an item basket, compared with its base year [3].

3. Representativeness: the basket must fit the group

• Basket: the fixed list of goods and services whose prices (or output) the index tracks.
• Weight: how important each item is in the index. It is usually the item's share in the group's spending (for CPI) or in value added (for IIP).
• The basket must match the reference group:
• A petrol price rise barely touches poor agricultural labourers.
• So petrol should carry little weight in their index.

• This is why India keeps separate CPIs: CPI-IW for industrial workers, CPI-AL/RL for agricultural and rural labourers, and CPI Rural/Urban/Combined for all households.

• How India now picks representative items:

• CPI 2024: the basket and weights come from the Household Consumption Expenditure Survey (HCES) 2023-24 [2].
• IIP 2022-23: manufacturing items come from the Annual Survey of Industries (ASI) 2021-22 and 2022-23. Items covering up to 90% of output in each industry group are chosen. The core basket must cover at least 80% of Gross Value of Output (GVO), and any item with more than 2% of an industry group's output is added so that new products are not missed [3].

• IIP sector weights = each sector's share in Gross Value Added (GVA) at current prices in 2022-23, taken from National Accounts with base 2022-23 [3].

• Representativeness gaps that were fixed:

• Rural housing: it had no weight ("NA") in CPI 2012. In CPI 2024 it carries 5.527 (group level, rural) [2].
• Online buying: CPI 2024 adds 12 online markets in 12 towns with more than 25 lakh people. Airfares, telephone charges and OTT subscriptions are also priced online [2].

• New sectors in IIP: it now covers gas supply, water supply, sewerage and waste management, minor minerals and a rare-earth mineral [3].

• What CPI leaves out: free social transfers, such as free grain, are excluded. By definition, CPI counts only items that households actually pay for. The IMF's CPI manual gives the same advice [2].

• Trap: free PDS grain does not enter CPI weights.

4. A "normal" base year

• Base year: the year whose prices anchor the comparison. The index is set to 100 in that year.

• Formula: Index in year t = (Value in year t ÷ Value in base year) × 100

• Rule 1: avoid extreme years such as droughts, wars or booms.

• Worked example:

  • Rice costs ₹20/kg in a normal year and ₹40/kg in a drought year. Today it costs ₹30/kg.
  • Drought year as base: (30 ÷ 40) × 100 = 75. Prices seem to have fallen.
  • Normal year as base: (30 ÷ 20) × 100 = 150. Prices have actually risen 50%.
  • A bad base year turns inflation into apparent deflation.

• Rule 2: the base should not be too distant.

• Comparing 1993 with 2005 is meaningful.
• Comparing 1960 with 2005 is not, because many items in the 1960 basket have since vanished.

• Real proof from the new IIP [3]:

  • 64 item groups dropped, e.g. kerosene, fluorescent tubes and CFLs, sewing machines, printing machinery.
  • 120 new item groups added, e.g. debit/credit cards, CCTV cameras, stents, vaccines, aircraft and spacecraft parts.

• Official test for IIP 2022-23: the base year should be "a relatively stable economic period" and should match the base year of GDP and WPI [3].

• Three "reference periods": CPI 2024 separates them. Examiners like this distinction [2].

Term	Meaning	CPI 2024 value
Index reference period	Period in which the index = 100	2024 = 100
Weight reference period	Period whose spending data give the weights	HCES 2023-24
Price reference period	Period in which base prices are collected	Calendar year 2024

5. Choice of formula

• The formula depends on the question being asked.
• Laspeyres index: uses base-year quantities (q₀) as weights.

• P_L = (Σp₁q₀ ÷ Σp₀q₀) × 100

• Paasche index: uses current-year quantities (q₁) as weights.

• P_P = (Σp₁q₁ ÷ Σp₀q₁) × 100

• Laspeyres and Paasche differ only in their weights.

• What India uses:
• IIP 2022-23: a Laspeyres fixed-base formula, L_t = ΣW_iR_i ÷ ΣW_i. Here W_i is the base-year weight of item i, and R_i = P_i,t ÷ P_i,0 is its production relative [3].
• CPI 2024 works in two steps [2]:
  • Jevons index (a geometric mean of price relatives) for elementary indices, i.e. the lowest level, where individual prices are combined.
  • Young / Modified Laspeyres index for higher-level indices.

• CPI 2024 structure: it adopts the UN's COICOP 2018 (Classification of Individual Consumption According to Purpose). This gives 12 Divisions, 43 Groups, 92 Classes and 162 Sub-classes, which makes India's CPI comparable with other countries [2].

• Why a fixed-weight index drifts over time:

• Laspeyres keeps old weights.
• When relative prices change, people switch to cheaper goods, but the index does not notice.
• So Laspeyres tends to overstate inflation. Paasche tends to understate it.
• This drift is one reason to revise the base regularly.

6. Reliable data

• Poor data mislead. Use primary data, or the most reliable secondary source.
• Improvements in CPI 2024 [2]:
• Prices are collected on tablets through CAPI (Computer Assisted Personal Interview) instead of paper.
• Coverage: 1,465 rural markets and 1,395 urban markets across 434 towns.
• Market prices are collected monthly. Online prices are collected weekly.
• Price collector: the Field Operations Division of NSS, MoSPI.
• Administrative data are used for some items:
  • rail fares from the Ministry of Railways;
  • fuel prices from PPAC (Ministry of Petroleum & Natural Gas);
  • postal charges from the Department of Posts;
  • mobile subscriber shares from TRAI.

• House rent is collected from 19,039 dwellings (15,715 urban, 3,324 rural).

• Data rules in IIP 2022-23 [3]:

• Closed or non-responding factories can be replaced by substitutes.
• If a factory does not respond for less than 6 months, its figure is estimated (imputation).
• WPI is used as the deflator for items reported in value terms. The Output Producer Price Index (PPI) will replace it once it proves stable.

7. Base-year revision: meaning and reasons

• Base-year revision: periodically moving an index's (or the national accounts') base year to a newer year. It keeps baskets, weights and data sources in line with what people consume and produce today.
• Why revise:
• Consumption patterns change.
  • As incomes rise, the share of food falls (Engel's law).
  • Spending on services, mobile data and packaged foods grows.
• Items drop out and new ones appear. Examples: kerosene and CFLs out; stents and CCTV cameras in [3].

• Better surveys and data sources become available, such as HCES 2023-24, ASI 2022-23, e-commerce prices and CAPI.

• Evidence: CPI weights, 2012 → 2024 (combined, old group structure) [2]

Group	CPI 2012	CPI 2024
Food and beverages	45.863	40.104
Housing	10.070	11.881
Fuel and light	6.843	5.489
Miscellaneous (services etc.)	28.317	33.154

• Under the new COICOP structure, food and beverages is 36.753 in CPI 2024. The 2012 series would have shown about 42.617 on the same structure [2].
• Transport rose from 6.394 to 8.796 (combined) [2].
• Weighted items rose from 299 to 358: goods from 259 to 308, services from 40 to 50 [2].

• Trap: "food weight fell to about 36.75%" (COICOP structure) and "fell to about 40.1%" (old structure) are both correct. The difference comes from classification, not from the data.

• Worked example: why new weights change the inflation number

• Assume food prices rise 10% and all other prices rise 4%.
• With old weights (food 45.86%): 0.4586×10 + 0.5414×4 = 4.59 + 2.17 = 6.75%
• With new weights (food 40.10%): 0.4010×10 + 0.5990×4 = 4.01 + 2.40 = 6.41%

• The same price changes give a lower CPI inflation figure, because food now has less weight.

• IIP basket, 2011-12 → 2022-23 [3]

• Total item groups: 407 → 463
• Manufacturing: 405 → 455
• Mining: 1 (29 minerals) → 3 (44 minerals)

• Electricity and gas: 1 → 3, with renewable electricity shown separately

• WPI 2022-23 [4][5]

• Items rose from 697 to 957.
• Solar, wind and nuclear electricity were added.
• Crude petroleum and natural gas moved from 'Primary Articles' to the 'Fuel and Power' group.
• New Producer Price Indices (PPIs) were launched: Output PPI, trial Input PPI, and Service PPI for 7 services (banking, securities, insurance, pension funds, railways, air passenger travel, telecom).

8. Splicing and the linking factor

• Splicing: joining an old series and a new series into one continuous series, using a linking factor, so that long-run comparison stays possible. DA calculations rely on these links.
• How MoSPI calculates it (CPI 2024) [2]:
• Both series are compiled for an overlap year. For CPI 2024 this is 2025.
• LF = Avg I_new ÷ Avg I_old, where each average is the geometric mean of the overlap year's monthly indices.
• Linked (back-series) index = I_old × LF
• Linking factors (general index): Rural 0.5222, Urban 0.5320, Combined 0.5267 [2].
• Worked example: an old-base (2012) CPI Combined reading of 190 becomes 190 × 0.5267 ≈ 100.1 on the 2024 base.

• A back series for Jan 2013 – Dec 2024 was published on 12 Feb 2026 [2].

• IIP: LF = GM of the old series during the new base year ÷ GM of the new series during the new base year. It is released for each sector [3].

• CPI-IW (2001 → 2016):
• The linking factor is 2.88 [7].
• The first index on the new base (September 2020) was 118. In old-base terms this is 118 × 2.88 ≈ 340 [7].
• Across the 65 centres common to both series, the factor ranged from 2.38 (Doom-Dooma Tinsukia) to 3.60 (Nagpur) [7].
• Direction trap:
  • The CPI-IW factor (2.88) turns a new index into its old-base value (new × LF).
  • The CPI 2024 factor (0.5267) turns an old index into its new-base value (old × LF).
  • Always check which way a linking factor works.

9. India's base-revision cycle (status as of Sep 2026)

Index	Compiler	NCERT base	Current / revised base
CPI-IW	Labour Bureau	2001	2016, released Oct 2020 (September 2020 index); LF 2.88 [7]
CPI Rural/Urban/Combined	NSO (MoSPI)	2012	2024 = 100, released 12 Feb 2026 [2][6]
WPI	OEA, DPIIT	2011-12	2022-23, approved 25 May 2026, released 15 Jun 2026 [4][5]
IIP	NSO (MoSPI)	2011-12	2022-23, planned release 1 Jun 2026; 10th base revision [3]
GDP	NSO (MoSPI)	2011-12	2022-23, new series scheduled for 27 Feb 2026 [6]
CPI-AL/RL	Labour Bureau	1986-87	2019 = 100, replaces the 1986-87 series [8]

• History of IIP bases: 1937 → 1946 → 1951 → 1956 → 1960 → 1970 → 1980-81 → 1993-94 → 2004-05 → 2011-12 → 2022-23. The move to 2022-23 is the 10th revision [3].
• CPI history: MoSPI's CPI Rural/Urban/Combined started in January 2011 on base 2010 = 100. It moved to 2012, and now to 2024 [2].
• Aim:
• Align the bases of GDP, CPI, WPI and IIP, so that real growth, deflators and inflation are measured on the same footing.
• Revise every few years, not every decade or more.
• MoSPI plans to revise the CPI base every 3-5 years, in line with global practice and depending on when HCES data become available [2].

Prelims Hooks

• Laspeyres vs Paasche: they differ only in weights. Laspeyres uses base-year quantities (q₀); Paasche uses current-year quantities (q₁).
• CPI 2024 formulas: Jevons (geometric mean) at the elementary level, and Young / Modified Laspeyres at higher levels. IIP uses a Laspeyres fixed-base formula [2][3].
• Weight sources: CPI 2024 weights come from HCES 2023-24. IIP 2022-23 sector weights come from GVA at current prices, 2022-23 [2][3].
• Food weight in CPI 2024 (combined, COICOP structure) = 36.75%. On the old structure it is 40.10%, down from 45.86% in CPI 2012 [2].
• Linking factors: CPI-IW 2001 → 2016 = 2.88. CPI 2012 → 2024 (combined) = 0.5267, calculated on the 2025 overlap year using geometric means [2][7].
• Free PDS grain is not in CPI. Free social transfers are excluded, in line with the IMF CPI manual [2].
• Size of the new baskets: IIP 2022-23 has 463 item groups (120 added, 64 dropped). WPI 2022-23 has 957 items (up from 697) [3][4].
• Compiler pairs: WPI is compiled by the Office of the Economic Adviser, DPIIT (Ministry of Commerce and Industry), not MoSPI. CPI-IW and CPI-AL/RL are compiled by the Labour Bureau. CPI (R/U/C), IIP and GDP are compiled by NSO, MoSPI.
• CPI-AL/RL now uses base 2019 = 100, replacing 1986-87 [8].
• Three CPI periods: index reference = 2024; weight reference = HCES 2023-24; price reference = calendar year 2024 [2].

Mains Points

• Old bases distort policy.
• An old CPI basket gave food about 46% weight.
• So every vegetable price spike pushed headline inflation up sharply.
• That affected RBI's flexible inflation targeting (4% ± 2%).

• Updated weights (food about 36.75-40%) and more services give a truer picture of household costs. The trade-off: inflation may now respond less to food shocks that still hurt poor households.

• Aligned bases improve real numbers.

• GDP, WPI, IIP and CPI all moving to 2022-23 / 2024 (2026) means deflators and real growth rest on the same economic structure.
• Earlier, old WPI weights used to deflate a service-heavy GDP could misstate real growth.

• A Service PPI is the long-term fix [3][5][6].

• A representative index needs both a regular cycle and better data.

• Revising every 3-5 years, linked to HCES rounds [2], is a GS-III theme on reforming India's statistical system.
• Other upgrades support it: CAPI, e-commerce prices, administrative data, and the COICOP 2018 structure that makes India's CPI comparable with other countries.

• These address earlier criticism of the credibility of Indian data.

• Splicing has distributional effects.

• DA for about 1 crore-plus central employees and pensioners, and wage indexation under labour codes, depend on CPI-IW and its linking factor.
• A wrong link or a biased basket directly changes real wages. So base revision is also a welfare and equity issue (GS-II/GS-III).