Income precarity links differently to India’s handloom sustainability outcomes

India’s handloom sector shows three distinct associations between local income precarity and sustainability-related outcomes, according to a cross-sectional study of administrative data from 32 handloom-producing regions. Geographies with a larger share of households earning below ₹5,000 per month had higher rates of cooperative participation, while their natural-dye adoption showed a negative but statistically imprecise association. Export and e-commerce accounted for a very small share of reported sales, and the income coefficients for that measure changed sign when individual regions were excluded.

The analysis concerns differences among rural and urban geography-segments, not individual household behaviour, and cannot establish that income composition caused any outcome. The study, published in SN Social Sciences, used geo-aggregated records from the National Data and Analytics Platform operated by NITI Aayog, Government of India. The source contained 64 rural or urban segments nested within 32 parent geographies.

Sixty-three observations had valid household denominators and were used in the cooperative-participation and natural-dye models, while 61 met a prespecified threshold of at least 20 reported sales and were used for the non-local sales model. A Delhi urban segment with no reported handloom households was retained in the master dataset but treated as missing for household-denominator measures. Because segments within a parent geography may share institutional, market, policy and reporting conditions, statistical uncertainty was estimated with CR2 cluster-robust covariance matrices clustered at the parent-geography level and Satterthwaite small-sample adjustments.

Income precarity was represented by the composition of households in three monthly-income groups: below ₹5,000, ₹5,001–₹10,000 and above ₹10,000. Household-weighted results showed that 67.08% of handloom households were in the lowest bracket, 26.20% earned ₹5,001–₹10,000 and 6.72% earned more than ₹10,000. Thus, 93.28% were in the two groups below ₹10,000 per month in the analytical sample.

Because these shares sum to 100%, a coefficient for either lower-income category represents a compositional reallocation from the omitted higher-income category, not an independent rise in income. The primary models were denominator-weighted quasibinomial logistic regressions, with household totals used to weight the cooperative and natural-dye models and reported sales observations used to weight the transaction model. Preliminary diagnostics showed overdispersion in all three models.

Cooperative engagement was defined as the number of cooperative members divided by the number of handloom households in each geography-segment. A 10-percentage-point compositional reallocation toward households earning below ₹5,000 was associated with higher odds of cooperative participation, with an odds ratio of 2.16 and a 95% confidence interval of 1.01–4.63; the cluster-adjusted p-value was 0.049. Holding other model variables at their sample medians, the model predicted cooperative participation of 2.46% at the 25th percentile of the below-₹5,000 share, 10.15% at the median and 30.78% at the 75th percentile, with wide confidence intervals, especially at the upper end.

The positive coefficient direction did not reverse in leave-one-segment-out or leave-one-parent-geography-out analyses, but its magnitude varied. The result therefore provides a directionally stable, geographically heterogeneous association rather than a strong household-level effect. Natural-dye adoption, defined as the share of handloom households reporting natural-dye use, showed the opposite directional pattern.

A 10-point compositional reallocation toward the below-₹5,000 category was associated with lower odds of adoption: the odds ratio was 0.61, the 95% confidence interval was 0.24–1.57 and p = 0.222. The model predicted adoption declining from 68.65% at the 25th percentile of low-income prevalence to 45.81% at the median and 26.22% at the 75th percentile, although cluster-robust intervals widened away from the centre of the distribution. The negative coefficient direction persisted across all influence exclusions, including removal of Andhra Pradesh or Mizoram, but the full-sample confidence interval included no association.

This pattern is consistent with the hypothesis that financially vulnerable producers may have less capacity to absorb adoption costs, technical uncertainty, quality-control demands or certification requirements, yet none of those mechanisms was directly measured. The third outcome had a different empirical meaning. Rather than estimating the proportion of households selling through export or e-commerce, Model C measured the share of all reported sales transactions occurring through those channels.

Its numerator combined export and e-commerce transactions, while its denominator included reported local, cooperative, export and e-commerce sales. The minimum threshold of 20 transactions excluded two segments with extremely small denominators. In the full sample, each 10-point compositional reallocation toward a lower-income category was positively associated with the non-local transaction share, and the confidence intervals were extremely narrow.

However, the absolute model-predicted shares were only 0.243% at the 25th percentile of the below-₹5,000 share, 0.481% at the median and 0.981% at the 75th percentile. That apparently precise market result was not robust to influence analysis. Both income-share coefficients changed sign when one geography-segment was removed at a time and when all observations belonging to one parent geography were excluded, with West Bengal and Andhra Pradesh producing the most extreme changes.

The study therefore did not support a dependable positive relationship between income precarity and market integration. The analysis did not directly observe logistics, product cataloguing, design support, digital infrastructure, certification, payment systems, intermediaries or buyer relationships. These factors may be relevant contextual conditions, but the models did not test them as explanations.

The low predicted transaction shares and unstable coefficient directions show that income composition alone provides an incomplete account of non-local sales, rather than demonstrating that poorer places achieve greater market integration. Across the analytical sample, the reported descriptive rates were 12.64% for cooperative participation and 43.69% for natural-dye use, while the sales-weighted export and e-commerce transaction share was 0.55%. A synthetic income index also indicated modest between-geography inequality: assigning ₹10,000 to every household in the top-coded category above ₹10,000 produced a household-weighted Gini coefficient of 0.080, with a 95% bootstrap confidence interval of 0.037–0.136.

Alternative upper-income assumptions produced estimates from 0.080 to 0.113. This measure describes dispersion among geographic averages, not inequality among households within a geography. The results carried substantial limitations.

Administrative counts could contain reporting, classification or measurement error, and the outcome categories were not mutually exclusive: one household could be a cooperative member and a natural-dye user, for example. Models A and B used household rates, whereas Model C used reported sales transactions, so their levels and denominators cannot be compared as equivalent participation probabilities. The geography units combined heterogeneous state, district or production-area boundaries with rural and urban segments.

Credit access, cooperative governance, certification, logistics quality, digital connectivity, design support, buyer networks and state-programme intensity were not observed. With a cross-sectional design, only 32 parent-geography clusters and a modest number of geography-segment observations, the analysis identifies suggestive between-place associations but cannot establish temporal order or identify causal effects. The models also adjusted for one cross-mechanism characteristic, but those controls should not be interpreted as established mediators.

The central finding is therefore not that one low-income condition produces a uniform sustainability response. Greater income precarity was associated with more cooperative participation and showed a directionally stable negative association with natural-dye adoption, although the latter remained statistically imprecise. The full-sample association with export and e-commerce transactions was positive but neither substantively large nor influence-robust.

These patterns are compatible with cooperative risk-sharing, financial constraints on ecological upgrading and the importance of institutional infrastructure, but the proposed mechanisms remain hypotheses. Direct testing would require household-level microdata, longitudinal observations that track changes within weaving clusters, or intervention-based designs that measure finance, certification, design support, logistics and market connections.

Subject of Research: Cross-sectional associations between income composition, cooperation, natural-dye adoption and non-local sales in India’s handloom sector

Article Title: Bridging livelihoods and sustainability in India’s handloom sector: linking income precarity, green adoption, and market integration

Article References: Choudhari, S. M. (2026). Bridging livelihoods and sustainability in India’s handloom sector: linking income precarity, green adoption, and market integration. SN Social Sciences, 6(11), Article 564. https://doi.org/10.1007/s43545-026-01866-y

Image Credits: AI Generated

DOI: 10.1007/s43545-026-01866-y

Keywords: Indian handloom sector, income precarity, cooperative participation, natural dyes, e-commerce, export sales, sustainable textiles, administrative data, cross-sectional study, Bridging, livelihoods, sustainability