Households: what households say, spend, pay with and borrow

The Households dimension asks “Are people spending, in cities and villages alike?” This note covers the data behind /economy/households/ and its sub-pages.

Code: pipeline/tipsheet/compute/economy_households.py; publish/economy_households.py (publish_households); step economy_households in steps/economy_households.py (needs macro, pulse, economy_growth; about 2 seconds). Tests: pipeline/tests/test_economy_households.py. Bundles: economy/households/confidence, perceptions, perceptions_latest, inflation_views, flows, bounce, credit, credit_latest, doca, hces, c85, facts. Pages: /economy/households/ (overview), mood/, cities-villages/, payments/, borrowing/ (apps/site/src/pages/economy/households/, helpers in apps/site/src/lib/economy-households.ts). Pre-registration and results: C85 in docs/research/site_calculations_spec.md.

What households say

Source. RBI’s survey workbooks in the Data bank (rbi_survey_tables): the urban Consumer Confidence Survey (uccs, 19 cities, bimonthly; quarterly before 2015), the rural Consumer Confidence Survey (rccs, from September 2023 in the workbook) and the Inflation Expectations Survey of Households (iesh). For each table, key and round the newest vintage is used; rounds are dated by survey month. Standard errors, which RBI prints beside some medians, are kept apart (measure = standard_error) and not mixed in.

• Net responses for the economy, employment, the price level, income, spending, essential and non-essential spending: the share saying better (more) less the share saying worse (less), as RBI prints it, now against a year ago and a year ahead. For the price level RBI’s sign is fell less rose.
• Indices. The current situation index (CSI) and future expectations index (FEI): 100 plus the average of the net responses on the economy, employment, prices (sign reversed), income and spending. The urban CSI and FEI come from the workbook where it is held (from March 2015) and from India Data Hub’s copy of the same RBI series (HHCCCURIDX11T, HHCCFUTIDX11T) for earlier rounds back to September 2012; in the 67 rounds both carry, they agree to 0.004. If the workbook dataset is missing, the hub series stand alone; rural readings then drop out.
• Inflation views. Urban: IESH median current and one-year-ahead inflation. Rural: the rural confidence survey’s own median perception and expectation (Table 10). Different surveys and samples: the page compares directions, not levels. The full treatment of expectations is on /economy/prices/expectations/ (C71).
• Licence. These are RBI’s own levels. Bundles carry meta.rbi_level, and pages chart them inside the no-download wrapper (no table, CSV or JSON).

What households spend and pay with

• Flows (flows): Pulse’s monthly readings (pulse_monthly.heat, growth on the same month a year earlier) for UPI person-to-merchant payments, credit and debit card spending, ATM cash, two-wheeler, car and tractor registrations, domestic air passengers and petrol. Sources and treatment as in docs/methods/pulse.md.
• Rural against urban. No monthly series splits spending by area. Two-wheelers and tractors stand in for rural demand and cars for urban; the overview compares their six-month averages of growth.
• Household consumption (PFCE) comes from the Growth step (economy_growth_quarterly, bases linked as on the Growth pages); the pre-pandemic average is April 2012 to December 2019 on the old base.
• HCES (hces): MoSPI’s Household Consumption Expenditure Survey 2022-23 and 2023-24 from /macro/‘s tables: average monthly per-capita spending (with imputation), the food share (“Food Total”), the bottom and top 5% (MoSPI’s fractile table, which is without imputation) and the Gini coefficient. The 2011-12 survey used a different design and is not set beside them.

What households borrow

• Loans to individuals (credit, credit_latest): RBI’s monthly sectoral deployment workbooks (rbi_sectoral_credit), the newest release for each reporting date, one date a month. Parts are matched by label, because RBI renumbered them in December 2021 (code 4.8 was “other personal loans”, then gold loans). Growth is RBI’s own, printed beside the year-ago amount in the same workbook, so it compares like with like; months whose workbook, as held, prints no growth stay blank (most of 2021 to 2023). “Other personal loans” is shown only from December 2021, when gold loans were split out of it. Housing is blank from July 2023 to June 2024 (HDFC’s merger). The headline uses the registry’s loader (economy_registry.sector_credit("personal")), which carries RBI’s merger-adjusted figure for that year and India Data Hub’s levels where the workbooks have no growth.
• Bounces (bounce): NPCI’s NACH statistics, recurring debits (debit_306__recurring_payments): returns as a share of presentations, by number and by value, monthly from July 2013. A return can be insufficient funds, a closed account or a mandate error.

Today’s bill and travel

• DoCA (doca): the Department of Consumer Affairs’ daily retail prices, all-India average, newest report. Each capture carries the price a month and a year earlier, so year-on-year change exists from the first day of capture (24 October 2026). One day’s prices, not an index.
• MoCA: the civil aviation ministry’s daily passenger dashboard is captured from 23 October 2026. It is not charted until a few weeks have built up; the pages say when capture began and point to DGCA’s monthly series.

Scorecard

facts.scorecard: each series’ place against its own history (economy_common.history_place), April 2020 to March 2022 left out of monthly series and Q1 FY21 to Q1 FY23 of PFCE. Short histories (the rural survey, UPI’s merchant split) are marked as new series.

C85: does urban confidence say anything about consumption?

Pre-registered before computing. The four-quarter change in the urban CSI (quarter’s mean of rounds) added to an AR(1) on last quarter’s real PFCE growth, forecast out of sample with an expanding window. Result (24 October 2026): 39 usable quarters, 23 evaluated; RMSE 1.72 against the AR(1)‘s 1.69 and the random walk’s 1.67; Diebold–Mariano p 0.71 (normal approximation, HLN correction). In-sample rank correlation of the change with PFCE growth 0.14, of the level 0.46. Does not pass; the pages say confidence is read for mood, not as a forecast.

Caveats

• The confidence surveys cover 19 large cities and, from 2023, villages; they ask which way things moved, not by how much.
• Vehicles count registrations, not buyers.
• Payment growth mixes more spending with a shift from cash.
• PFCE is partly a residual in the national accounts and is revised.