Consumption surveys: HCES, MPCE and reference periods

Detail

1. What a consumption survey is

• Consumer expenditure survey (now called the Household Consumption Expenditure Survey, HCES): a sample survey. Chosen households are asked what they bought and used over a set period.
• It is conducted by the National Sample Survey (NSS), which comes under MoSPI (Ministry of Statistics and Programme Implementation).
• MoSPI states its purpose: HCES data is used to judge trends in economic well-being and to update the basket and weights of the Consumer Price Index (CPI). It is also used to measure poverty, inequality and social exclusion [2].
• MoSPI calls MPCE "the primary indicator used for most analytical purposes" [2].

2. History: from the 1st round to the missing 75th round

• 1st round (1950-51): the first NSS consumer expenditure survey.
• "Thick" rounds: large-sample surveys held about every 5 years (quinquennial). They ran from the 27th round (1972-73) to the 68th round (2011-12). The small yearly surveys in between were called "thin" rounds.
• The 68th round's fieldwork ran July 2011 to June 2012 [2].

• NCERT uses the 68th round as its example. This is now outdated (NCERT outdated).

• 75th round (2017-18): it was withheld in November 2019. The government cited "data quality". This came after a leaked report showed real MPCE falling (spending falling after inflation is removed).

• Result: India had no official consumption data from 2011-12 until 2022-23. For about 11 years there was no official poverty estimate and no fresh set of CPI weights.

• After Covid, MoSPI decided to hold two surveys back to back, in 2022-23 and 2023-24 [2].

• HCES 2022-23 factsheet: February 2024. Detailed report and unit-level data: June 2024 [2].
• HCES 2023-24 factsheet: 27 December 2024 [2].

3. HCES 2022-23 and 2023-24: how the survey is designed

• Survey periods: August 2022 to July 2023 and August 2023 to July 2024 [2].
• Sample size (2023-24): 2,61,953 households. Of these, 1,54,357 were rural and 1,07,596 were urban, covering all States and UTs [2].

• Three questionnaires: 1. food; 2. consumables and services; 3. durables (goods that last a long time, like fridges or motorcycles).

• Three monthly visits: each questionnaire was filled in on a separate visit, over three months.

• Why: in one long interview, people get tired and give careless answers. This is called respondent fatigue, and it is a non-sampling error. Splitting the interview across visits reduces it.

• CAPI (Computer Assisted Personal Interview): the field worker enters answers on a tablet instead of paper. The tablet checks answers straight away, so there are fewer entry errors.

4. MPCE: definition and worked example

• Monthly Per Capita Consumption Expenditure (MPCE): a household's average spending per person per month.

MPCE = Household's monthly consumption expenditure ÷ Household size

• Worked example:
• A rural household of 5 people spends Rs 20,000 in a month.
• MPCE = 20,000 ÷ 5 = Rs 4,000.

• Every member of this household gets the same MPCE, Rs 4,000. Households, and the people in them, are then ranked by MPCE.

• Two sets of MPCE, reported in both 2022-23 and 2023-24 [2]:

• (i) Without imputation: only what the household spent, plus the usual imputed value of home-grown produce, gifts, loans and free collection.
• (ii) With imputation: also adds a money value for items received free through welfare schemes [2]:
  • Food: rice, wheat/atta, jowar, bajra, maize, ragi, barley, small millets, pulses, gram, salt, sugar, edible oil (for example, PDS grain);
  • Non-food: laptop/PC, tablet, mobile handset, bicycle, motorcycle/scooty, school uniform, school shoes.

• Imputation means giving an estimated money value to something the household did not pay for.

• Not imputed: free health care under PM-JAY (cashless treatment, where the household does not know the bill) and free education (fee waivers or reimbursement). MoSPI said valuing these was too difficult [2].

• Worked example (2023-24): rural MPCE rises from Rs 4,122 to Rs 4,247 after imputation. That is Rs 125 more, or about 3%. Urban MPCE rises from Rs 6,996 to Rs 7,078 [2].
• Why the rural rise is bigger: free transfers such as PDS grain matter more to rural households.

5. Results

	2011-12 (68th round)	2022-23	2023-24
Rural MPCE (current prices)	Rs 1,430 [2]	Rs 3,773	Rs 4,122
Urban MPCE (current prices)	Rs 2,630 [2]	Rs 6,459	Rs 6,996
Rural MPCE at 2011-12 prices	Rs 1,430	Rs 2,008 [2]	Rs 2,079 [2]
Urban MPCE at 2011-12 prices	Rs 2,630	Rs 3,510 [2]	Rs 3,632 [2]
Urban-rural gap	84%	71%	70%
Gini: rural / urban	—	0.266 / 0.314	0.237 / 0.284

• Nominal growth (2022-23 to 2023-24): about 9% rural and 8% urban. "Nominal" means before removing inflation [2].
• Urban-rural gap: worked example (2023-24)
• Gap = (Urban MPCE − Rural MPCE) ÷ Rural MPCE × 100
• = (6,996 − 4,122) ÷ 4,122 × 100 ≈ 70%
• Meaning: an average urban person spends about 1.7 times what an average rural person spends.

• States (2023-24): the gap is lowest in Kerala (about 18%) and highest in Jharkhand (about 83%) [4].

• Gini coefficient: a measure of inequality. 0 means everyone spends the same. 1 means one person does all the spending. A falling Gini means consumption is spread more evenly.

• Range across States (2023-24):
• Highest MPCE: Sikkim, Rs 9,377 rural and Rs 13,927 urban.

• Lowest MPCE: Chhattisgarh, Rs 2,739 rural and Rs 4,927 urban [4].

• Food share below 50%: food is about 47% of rural and about 40% of urban MPCE (2023-24). Non-food is about 53% rural and 60% urban [2].

• Engel's law: as people get richer, the share of income spent on food falls. A food share below half is a sign that living standards are rising.

• Largest food item: beverages, refreshments and processed food, at 9.84% of rural MPCE. Next come milk and milk products (8.44%) and vegetables (6.03%), all rural 2023-24 [4].

• Largest non-food items: conveyance (travel), clothing, bedding and footwear, entertainment, and durables. In urban households, rent is about 7% [2].

6. Fractile classes: who gains

• Fractile class: the population is ranked by MPCE and cut into groups, from the bottom 5% up to the top 5% (see §6, percentiles).
• MoSPI's definition: for a fraction f, the fractile Yf is the MPCE level below which a share f of people fall [2].

• Example: the 0.05 fractile is the MPCE below which the poorest 5% fall.

• 2023-24 figures [2]:

Fractile class	Rural MPCE	Urban MPCE
Bottom 0-5%	Rs 1,677	Rs 2,376
Top 95-100%	Rs 10,137	Rs 20,310

• Top-to-bottom ratio: about 6 times in rural India (10,137 ÷ 1,677) and about 8.5 times in urban India (20,310 ÷ 2,376).
• Bottom 5% grew fastest (2022-23 to 2023-24): 22% rural and about 19% urban [2].
• Why it matters: the poorest gained the most, and this matches the fall in the Gini.

7. Reference periods (recall periods)

• Reference (recall) period: the length of past time a household is asked to remember its spending, for example "in the last 30 days".

Method	Recall used
Uniform reference period (URP)	30 days for all items
Mixed reference period (MRP)	365 days for clothing, bedding, footwear, durables, education and institutional medical care; 30 days for all else
Modified mixed reference period (MMRP) (from 2009-10)	7 days for perishables (edible oil, egg-fish-meat, vegetables, fruits, spices, beverages, processed food, pan/tobacco/intoxicants); 365 days for the MRP items; 30 days for the rest

• Logic of each period:
• Rarely bought items (a fridge, school fees) → 365 days. A 30-day window would miss most of these purchases.

• Items bought every day (vegetables, milk tea) → 7 days. People forget small daily purchases over a whole month.

• The 2011-12 and HCES figures in §5 are on the MMRP basis.

8. Why recall matters

• A shorter recall captures more spending, because people forget less.
• Small daily purchases are quickly forgotten.
• A 7-day question captures more of them than a 30-day question.

• So measured spending goes up.

• Therefore: MPCE under MMRP > MRP > URP.

• Effect on poverty:
• Higher measured MPCE → more people are above the poverty line.
• So measured poverty is lowest under MMRP.

• Actual living standards have not changed; only the method has.

• This is a non-sampling error (§4) changing a headline number.

• Non-sampling error: an error that comes from how data is collected (memory, question wording, refusals). It does not come from surveying a sample instead of everyone.
• A bigger sample does not fix it.

9. One survey, three uses

(a) Poverty lines (analysis → poverty-inequality)

• Lakdawala method used URP.
• Tendulkar method used MRP.
• Rangarajan method used MMRP.
• Exam trap: poverty numbers from different committees cannot be compared directly, partly because their recall methods differ.

(b) CPI weights (construction → inflation-price-indices)

• Old series: 2011-12 MMRP data gave the weights for CPI (base 2012).
• New series: CPI base 2024 = 100 is now live, with weights taken from HCES 2023-24 [3]. (The scaffold said "verify current": now confirmed.)
• Weight reference period: HCES 2023-24.
• Price reference period: calendar year 2024 [3].

• The back series linked to base 2024 was set for publication on 12 February 2026 [3].

• Weight of food and beverages [3]:

• Under the old classification, it would fall from 45.86% (CPI 2012) to 40.10% (CPI 2024).
• Under the new COICOP 2018 structure, it is 36.75% (CPI 2024), against about 42.62% for the 2012 series.

• COICOP (Classification of Individual Consumption According to Purpose): the UN Statistics Division's system for grouping household spending.

• Size of the basket: items rose from 299 to 358; services went from 40 to 50 [3].

• Free items are excluded from CPI weights: a CPI should count only what households actually pay for. The IMF CPI manual advises leaving out social transfers in kind [3].
• So the CPI uses spending without the imputed value of free items.

(c) Cross-check on PFCE

• PFCE (Private Final Consumption Expenditure): total household spending as estimated in the National Accounts (NAS).
• Survey consumption is compared with PFCE. The survey total usually comes out lower, and this is called the NSS-NAS gap.
• Causes:
• rich households often refuse to answer (non-response);
• those who do answer under-report their spending;

• the two use different definitions and coverage.

• Effect: if the rich are missed, the survey can also understate inequality. The Gini may look lower than it truly is.

Prelims Hooks

• MPCE = household monthly consumption expenditure ÷ household size. It is the main HCES indicator [2].
• HCES 2023-24: rural MPCE Rs 4,122 and urban Rs 6,996. With imputed free items: Rs 4,247 and Rs 7,078 [2].
• Urban-rural MPCE gap: 84% (2011-12) → 71% (2022-23) → 70% (2023-24) [2].
• Gini (2023-24): rural 0.237 and urban 0.284. Both fell from 2022-23.
• MMRP (from 2009-10) uses a 7-day recall for perishables. URP uses 30 days for everything. Order of MPCE: MMRP > MRP > URP.
• Committee–method pairs: Lakdawala–URP, Tendulkar–MRP, Rangarajan–MMRP.
• CPI (base 2024) takes its weights from HCES 2023-24 and follows COICOP 2018. Free social transfers are excluded [3].
• Trap: free health care (PM-JAY) and fee waivers were not imputed in HCES MPCE. Only listed goods such as PDS grain, laptops and bicycles were [2].
• Trap: the 75th round (2017-18) was withheld in 2019. The official series jumps from 2011-12 straight to 2022-23.
• Largest food item in MPCE: beverages, refreshments and processed food, not cereals [4].

Mains Points

• Method shapes the number:
• Recall period (URP/MRP/MMRP) and imputation of free items change MPCE, and so change poverty estimates.
• When comparing across years, the method must be kept the same.

• Headline "poverty fell" claims should state the method used.

• Credibility of statistics:

• Withholding the 75th round left an 11-year data gap.
• This delayed CPI weight revision and poverty measurement.

• It supports calls for an independent statistical system with a fixed release calendar.

• Welfare reading of HCES 2023-24:

• Food share is below 50%, the rural-urban gap is narrowing, the Gini is falling, and the bottom 5% grew fastest [2].
• Together these suggest a rise in rural living standards.

• Caution: under-reporting by the rich and the NSS-PFCE gap may understate inequality.

• Policy linkages:

• Fresh HCES weights give a CPI 2024 basket with a smaller food weight [3].
• The RBI targets CPI inflation, so this changes how food-price shocks show up in headline inflation and in inflation targeting.