Technical Methodology - India Political Economy Assessment

1. Methodological Overview

This study employs a mixed-methods approach combining quantitative analysis of official statistics, independent surveys, and international assessments with qualitative analysis of policy documents and institutional changes. The methodology is designed to address the challenge of analyzing an economy where official data reliability is itself a research question.

Core Methodological Principles

Triangulation: No single data source is treated as definitive; findings emerge from convergent evidence
Transparency: All data sources, transformations, and code are publicly available
Conservatism: When data conflict, we report ranges and note uncertainties
Replicability: All analyses can be reproduced using provided code and data

2. Data Sources

2.1 Government Sources

National Statistical Office (NSO)

GDP and National Accounts (2004–2026)
Consumer Price Index series
Industrial Production Index
Annual Survey of Industries (ASI)

Reserve Bank of India (RBI)

State Finances: Annual Studies (2004-2024)
Monetary Policy Reports
Financial Stability Reports
Household Finance Committee data

Labour Force Surveys

Employment-Unemployment Surveys (2004-05, 2009-10, 2011-12)
Periodic Labour Force Survey (2017-18 to 2023-24)
Quarterly Employment Surveys (urban)
EPFO/ESIC payroll data (2017-2024)

2.2 Independent Sources

Centre for Monitoring Indian Economy (CMIE)

Consumer Pyramids Household Survey
Employment-Unemployment data
Household Income and Expenditure
Coverage: 174,405 households, quarterly

International Databases

World Inequality Database (WID)
V-Dem Institute Dataset v14
Freedom House Annual Reports
Reporters Without Borders Press Freedom Index
IMF Article IV Consultations
World Bank Development Indicators

3. Novel Indices Construction

3.1 Statistical Suppression Index (SSI)

The SSI is a weighted sum of graded suppression severities (0–10 scale), with a persistence/scar term so a resolved suppression decays rather than dropping to zero. Computed identically for both eras in data/compute_indices.py; the UPA decade records nothing, so SSI = 0.

SSI_t = Σ(w_i × s_it) where: s_it = severity of stream i in year t ∈ [0,1] (1.0 full suppression/discontinuation; ~0.7 methodology dispute; ~0.3 delay or post-resolution scar) w_i = weight of stream i (see table; max total = 10)

Component	Weight	Trigger / event
Census delay	2.5	2021 census postponed beyond constitutional mandate
Consumption survey	1.5	2017-18 survey withheld; 11-year data gap
Employment data	1.0	PLFS 2017-18 withheld until after 2019 election
GDP back-series	1.5	2011-12 base + shelved 2018 back-series, unreviewed
Institutional independence	1.5	NSC resignations (2019) + permanent NSSO→NSO merger
Mortality undercount	2.0	COVID deaths: official ~0.5M vs 3–5M excess-mortality estimates

3.2 Fiscal Centralisation Index (FCI)

The FCI measures erosion of fiscal federalism — six components, each min-max normalised over the full 2004–2026 sample, then averaged:

FCI_t = (1/6) × Σ[C_jt] Components: C1: Cess/Surcharge share (normalized) C2: Effective devolution (inverted) C3: States' own revenue capacity (inverted) C4: CSS conditionality C5: Borrowing restrictions C6: GST structural centralisation (0 pre-2017; full once GST compensation ended mid-2022)

3.3 Democratic Quality Index (DQI)

The DQI uses a geometric mean of four published third-party measures to penalize weakness in any dimension:

DQI_t = (V_t × F_t × R_t × C_t)^(1/4) where: V_t = V-Dem Liberal Democracy Index F_t = Freedom House Score (normalized) R_t = RSF Press Freedom (inverted rank) C_t = V-Dem Core Civil Society Index

4. Analytical Methods

4.1 Difference-in-Differences Analysis

For policy impact assessment:

Y_it = α + β(Period_t × Treatment_i) + γX_it + δ_i + θ_t + ε_it

Applied to GST implementation, demonetization, and pandemic policies

4.2 Synthetic Control Method

Used for counterfactual analysis comparing India's trajectory with weighted combination of similar economies. Control units selected based on pre-2014 characteristics:

GDP per capita
Democratic institutions
Population size
Economic structure

4.3 Structural Break Tests

Test	Application	Break Points Identified
Chow Test	GDP series	2015 Q1, 2020 Q2
Bai-Perron	Employment	2016 Q4, 2020 Q1
Andrews-Ploberger	Inequality	2014 Q3

5. Triangulation Strategy

5.1 Cross-Validation Matrix

Each finding must be supported by at least three independent sources:

Example: Unemployment Rate Verification

PLFS: 6.1% (2017-18)
CMIE: 7.4% (2017-18)
Labour Bureau: 5.0% (2015-16)
Converged Estimate: 5.5-7.0% range

5.2 Dealing with Suppressed Data

Consumption Survey 2017-18: Reconstructed using CMIE household data + state consumption patterns
Census 2021: Projected using NFHS-5, voter rolls, and Aadhaar statistics
Caste Data: Extrapolated from state surveys (Bihar, Karnataka)

6. Limitations and Caveats

Critical Limitations

Census Gap: No census since 2011 limits demographic analysis
Consumption Data: 2017-18 survey suppressed; using imputations
State Variation: National aggregates mask significant state-level differences
Informal Economy: Measurement challenges persist despite methodology improvements
COVID Impact: Pandemic effects complicate 2020-2021 comparisons
Political Sensitivity: Some data sources may face pressure
The Production Boundary: Every GDP figure used here excludes unpaid household work. See below.

6.1 What GDP does not count

GDP measures work inside a production boundary set by the System of National Accounts. The boundary is a convention. Where a household owns the dwelling it occupies, the accounts estimate the rent it would otherwise have paid and add that to output, because a real service is being consumed. Cooking a meal in that dwelling, minding a child, nursing a parent: the same reasoning is not applied. Those services enter the accounts at zero.

India already collects the hours. The NSO ran Time Use Surveys through 2019 and 2024. Among those aged 15–59 who do it at all, women spent 289 minutes a day on unpaid domestic services in 2024 and men 88 — roughly 1,758 hours a year for a woman, in no employment statistic and no output series. Published valuations of that work span 7.5% to 36% of GDP, a spread driven by the choice of valuation method rather than by uncertainty about the hours.

No figure in this assessment is adjusted for it. Every growth rate is the official one on the official boundary. Full treatment: someperspective.info/care and Section 6.11 of the research paper.

7. Validation and Robustness

7.1 Sensitivity Analysis

All indices recalculated with alternative weights (±20%)
Results robust to exclusion of any single data source
Bootstrap confidence intervals for all estimates
Monte Carlo simulations for projection uncertainty

7.2 External Validation

Finding	External Validation	Correlation
Democratic erosion	V-Dem, Freedom House	0.89
Inequality rise	World Inequality Database	0.92
Employment crisis	ILO estimates	0.85
Fiscal centralization	Finance Commission	0.94

8. Replication Guide

8.1 Data Access

Repository: github.com/Varnasr/someperspective

Contents:

/data/raw/ - Original datasets
/data/processed/ - Cleaned data
/code/ - Analysis scripts (R, Python, Stata)
/docs/ - Codebooks and documentation

8.2 Software Requirements

R (v4.3+): Main statistical analysis
Python (v3.9+): Data processing and visualization
Stata (v17): Robustness checks
PostgreSQL: Database management

8.3 Reproduction Steps

Clone repository: git clone https://github.com/Varnasr/someperspective
Install dependencies: pip install -r requirements.txt
Run data processing: python process_data.py
Execute analysis: Rscript main_analysis.R
Generate figures: python create_figures.py

Technical Methodology Document | India Political Economy Assessment
Full research and data: someperspective.info
Last updated: May 2026