Urban–Rural Consumption Gap in India: A State-Level Regression Study of NSS 2011–12 and HCES 2022–23 Data

The Urban–Rural Consumption Gap in India is a secondary-data dissertation that studies why average household consumption differs so sharply between rural and urban India, and whether that gap has narrowed or persisted over the decade between the two big official consumption surveys.

The numbers are stark and official: the National Sample Survey's 68th round recorded average rural monthly per capita expenditure (MPCE) of about ₹1,430 in 2011–12, and MoSPI's Household Consumption Expenditure Survey 2022–23 puts rural MPCE at ₹3,773 against ₹6,459 in urban areas. Behind the national averages, however, states differ enormously — in the level of consumption, in the urban–rural ratio, and in how consumption is distributed across fractile classes.

The dissertation assembles a state-level panel from published MoSPI factsheets and the RBI Handbook of Statistics on Indian States, computes urban–rural gaps and Gini-style inequality measures from fractile shares, and then estimates cross-state regressions of average MPCE (and of the urban–rural gap) on structural determinants such as per-capita net state domestic product, urbanisation, schooling and non-farm employment shares — once for 2011–12 and once for 2022–23, so the two rounds can be compared.

Everything runs in R with Excel used for table preparation, which matches the Research Methodology and econometrics training of the UGCF Economics dissertation track. The final deliverables — a 12-credit dissertation report across Semesters 7 and 8, monthly progress reports, reproducible R scripts and a viva-voce defence — make it a rigorous but entirely feasible Honours-with-Research project.

India's official consumption statistics returned after an eleven-year gap when MoSPI published the HCES 2022–23 factsheet: rural average MPCE of ₹3,773 against urban ₹6,459. Public discussion, however, rarely moves past the national headline. We do not have a systematic, up-to-date account of how the urban–rural consumption gap varies across states, whether it narrowed or widened since NSS 2011–12, and which structural characteristics of states — per-capita income, urbanisation, education, non-farm employment — are associated with higher consumption and smaller gaps. Without that account, state-level welfare comparisons and policy discussion rest on impressions rather than measured evidence. The need is for a defensible, reproducible, state-level quantitative study that uses only published official data, applies the regression and diagnostics tools of the undergraduate econometrics syllabus, and states clearly the limits of cross-state correlational inference.

Abstract

Household consumption is the most direct welfare measure Indian official statistics offer, and the rural–urban consumption gap is one of its oldest stylised facts. This dissertation quantifies that gap state by state, measures how it changed between NSS 2011–12 and HCES 2022–23, and estimates what structural characteristics of states — income, urbanisation, education, employment structure — are associated with higher average consumption and narrower gaps. All analysis uses published official statistics, regression in R, and a pre-specified specification search recorded in a methodology log.

Introduction

In 2024 MoSPI released the factsheet of the Household Consumption Expenditure Survey 2022–23, the first official consumption round since 2011–12, reporting average MPCE of ₹3,773 in rural India and ₹6,459 in urban India. The same release allows comparisons with the 68th-round NSS figures of 2011–12 (rural about ₹1,430). For an economics undergraduate, this is a rare gift: two comparable rounds, eleven years apart, with state-level detail and fractile-class distributions published as open factsheets. The dissertation treats those releases as its laboratory.

Existing understanding and the gap

The literature on consumption inequality in India is large — National Sample Survey rounds have fuelled decades of work on poverty, Engel curves and the urban–rural divide — but almost all of it stops at 2011–12, because no newer consumption microdata was released for over a decade. The 2022–23 and 2023–24 HCES factsheets changed that, yet most public commentary stops at national averages and headline growth rates. The gap: a systematic, state-level, two-round comparison that (a) measures how the urban–rural gap moved in each major state, (b) relates those levels to state structural characteristics in a regression framework a UG student can defend, and (c) is fully honest about what cross-state correlations can and cannot establish.

Proposed study

• Data: HCES 2022–23 factsheet (state MPCE, fractile-class shares, food shares); NSS 68th-round key indicators for 2011–12; RBI Handbook of Statistics on Indian States for per-capita NSDP, urbanisation and schooling; PLFS summaries where an employment-share variable is needed.
• Measurement: urban–rural MPCE ratio and absolute gap per state; inequality approximated from published fractile shares (e.g. a Gini approximation); nominal-to-real conversion issues handled explicitly.
• Modelling: cross-state OLS of (log) MPCE and of the urban–rural gap on structural covariates, run separately for the two rounds, with robustness checks (outlier states, alternative specifications, population-weighted variants).
• Deliverables: reproducible R scripts, a two-round comparison atlas of figures, the dissertation, and a viva-ready defence of every specification.

Feasibility

• Technical: every input is a downloadable official table; the regression work is standard lm() diagnostics a UGCF econometrics course prepares students for.
• Economic: zero cost — no proprietary data or software is required.
• Operational: the 12-credit UGCF dissertation runs across Semesters 7 and 8 with monthly progress reports, which the 24-week plan maps directly; the main risk (figure comparability between NSS and HCES) is handled by an explicit comparability section rather than ignored.

In scope

• Major Indian states covered consistently by both NSS 2011–12 and HCES 2022–23 published tables.
• Secondary data only: MoSPI factsheets and reports, RBI Handbook of Statistics on Indian States, published PLFS summaries.
• Outcome variables: average MPCE (nominal and real), urban–rural MPCE ratio, fractile-share-based inequality approximation.
• Explanatory variables: per-capita NSDP, urbanisation share, mean schooling, non-farm employment share, and comparable structural indicators.
• Methods: descriptive comparison, Gini approximation, cross-state OLS with standard diagnostics in R.

Out of scope

• Unit-level microdata re-weighting or survey-design estimation (the study uses published tables).
• Causal identification — the design is explicitly correlational and says so.
• Sub-state (district) analysis, since HCES factsheets report state-level detail.
• Price-index construction from first principles; official/state indices and a documented nominal-vs-real discussion are used instead.

Research design: quantitative, non-experimental, cross-state comparative study over two survey rounds (2011–12 and 2022–23), using secondary data only. Because there are no human participants — every figure comes from published official tables — the study needs no ethics clearance, but it follows a written data-documentation protocol so every number can be traced to a source table and download date.

Instruments and analysis plan: R (tidyverse for data handling, ggplot2 for figures, lm() for estimation); Excel for initial table transcription with double-entry checks. Sampling is a census of major states (all states with comparable published tables in both rounds — around 20), not a convenience subset; every exclusion (e.g. states with unresolved comparability) is listed with reasons.

The study is an analysis pipeline rather than a software product, and its integrity depends on the flow below: each box is a file or script, each arrow an output consumed downstream.

flowchart TD
  A["MoSPI HCES 2022–23 factsheet and report tables"] --> D["Master state panel (CSV)"]
  B["NSS 68th-round 2011–12 key indicators"] --> D
  C["RBI Handbook of Statistics on Indian States"] --> D
  D --> E["Comparability notes: survey design, imputation, price basis"]
  E --> F["Measurement: urban–rural gap, real MPCE, Gini approximation"]
  F --> G["Exploratory atlas: maps, scatter plots, Lorenz curves"]
  F --> H["Cross-state OLS — 2011–12 round"]
  F --> I["Cross-state OLS — 2022–23 round"]
  H --> J["Diagnostics and robustness log"]
  I --> J
  G --> K["Dissertation: measurement chapter"]
  J --> L["Dissertation: results and interpretation chapters"]

Three design choices matter. First, the master panel stores a source note (publication, table number, download date) for every variable, which turns the viva question "where does this number come from?" into a lookup rather than a scramble. Second, the comparability notes box sits between data assembly and measurement because the NSS-to-HCES break (change in survey methodology and imputation between rounds) is the study's biggest threat to validity and must be handled explicitly, not in a footnote. Third, the two rounds are estimated as parallel regressions rather than pooled, since the methodology break would otherwise contaminate any pooled estimate; the diagram makes that parallel structure impossible to overlook.

• Add the HCES 2023–24 round for a three-point panel and test whether the gap's trajectory is monotone.
• Obtain HCES unit-level microdata when released and re-estimate with proper survey-design standard errors.
• Add district-level covariates (night lights, road density) to probe mechanisms behind state averages.
• Link consumption gaps to complementary welfare indicators — NFHS-5 nutrition and housing quality — for a multidimensional picture.
• Turn the measurement code into a small R package so later batches can update the panel with each new survey round.