Projects

Abstract

Multilateral negotiations at the UN General Assembly operate under extreme information asymmetry: delegations routinely receive revised draft text with hours to respond, while coalition signals shift through informal channels that resist structured capture. This project developed a lightweight negotiation operating system — a real-time tracker that mapped every active draft paragraph to the delegations with material interest, their stated and inferred positions, and the concessions they were likely to accept. Paired with a structured briefing workflow, hallway intelligence was converted into one-page decision memos within 30 minutes, enabling India's leadership to act on fresh signal while competing delegations were still convening internal reviews.

Key Findings & Impact

• Compressed India's internal decision cycle from 24–48 hours to under 4 hours on contested paragraphs, a 6–12x improvement
• Moved 6 multilateral resolutions (NAM, WSIS+20, ICZM, AgTech) from initial draft to adoption with India's core positions preserved
• Reduced surprise amendments by systematically mapping bloc-level and lobby-group signals to specific textual provisions
• Adopted as the de facto briefing standard within the Mission for Second Committee and technology-governance tracks

Methodology

Paragraph-level stakeholder mapping combined with a structured intelligence funnel: raw hallway signals were triaged against a position matrix, scored by reliability and urgency, and surfaced as decision-ready briefs. The system prioritised speed and actionability over comprehensiveness, reflecting the core insight that in multilateral settings, decision latency — not analytical depth — is the binding constraint.

Abstract

Landlocked developing countries (LLDCs) face compounding structural penalties: distance from ports, weak transit infrastructure, and thin institutional capacity to negotiate corridor access. Despite this, development monitoring for the 32 LLDCs relied on fragmented public data and narrative reports with limited cross-country comparability. This project constructed a unified analytical framework — consolidating 80+ macroeconomic, trade, infrastructure, and climate-vulnerability indicators into a policy handbook and interactive dashboard system — from over 50 stakeholder interviews and primary data reconciliation conducted over two weeks. Gravity model estimations using R quantified the trade penalties attributable to infrastructure gaps, providing the evidentiary basis that supported a successful lobbying effort securing approximately $3M from advanced economies in the EU.

Key Findings & Impact

• Built a comprehensive macroeconomic handbook covering 32 LLDCs with 80+ harmonised indicators, the first unified reference of its kind for UN-OHRLLS
• Gravity model regressions estimated that infrastructure deficits account for 18–34% of trade cost penalties in landlocked corridors
• Reported directly to the leadership of 32 countries (PM/President level) at the UN LDC5 preparatory process
• Analytical outputs contributed to the lobbying case that mobilised ~$3M in commitments from EU member states
• Completed as the youngest economist among peer analysts from leading global institutions

Methodology

Cross-national indicator harmonisation from World Bank, UNCTAD, and national statistical office sources, cleaned and reconciled through semi-structured stakeholder interviews. Trade penalty estimation via augmented gravity models (OLS and PPML) in R, controlling for bilateral distance, colonial ties, common language, and infrastructure quality indices. Dashboard construction in Python for real-time monitoring.

Abstract

Anti-dumping enforcement in the U.S. steel sector has historically been reactive: investigations are opened after domestic producers file petitions, by which point circumvention routes are established and market damage is done. This project developed an AI-based early warning framework that detects circumvention risk before it materialises, using anomaly detection on HS-6 level export flows, firm-level shipment patterns, and supply-demand gap indicators across the steel value chain. The system flagged transshipment risks through third-country routing (principally Vietnam, Malaysia, and Thailand) and generated risk scores calibrated against historical Commerce Department determinations. By shifting enforcement from reactive petition-response to proactive risk identification, the framework improved monitoring efficiency by approximately 25% by Q2 2024.

Key Findings & Impact

• Identified statistically significant anomalies in HS-6 steel export flows from 3 key transshipment countries, predicting 78% of subsequently confirmed circumvention cases in the validation set
• Firm-level analysis revealed that 12% of exporters in flagged corridors accounted for 60%+ of anomalous volume patterns
• Supply-demand gap model detected capacity mismatches 4–6 months before traditional indicators, enabling preemptive review
• Improved monitoring efficiency by ~25%, reducing the analyst-hours required per enforcement case
• Presented to Commerce Department officials; framework designed for integration with existing ITA workflows

Methodology

Anomaly detection on bilateral trade flows (UN Comtrade, USITC DataWeb) using STATA and R. Firm-level shipment analysis using customs microdata. Supply-demand gap estimation via capacity utilisation models. Risk scoring calibrated against historical anti-dumping and countervailing duty determinations. Classification framework validated against out-of-sample enforcement actions.

Abstract

India's AI governance portfolio at the United Nations spans multiple negotiation tracks — from the Global Digital Compact to WSIS+20 — each involving distinct coalitions, redline issues, and rapidly evolving draft text. Traditional briefing methods could not keep pace with the volume of parallel processes. This project built a suite of AI-driven tools: 6 geopolitical dashboards monitoring delegation positions across active negotiation tracks, an automated briefing pipeline that synthesised overnight developments into structured morning reports, and a negotiation tracker correlating textual amendments with bloc-level strategy signals. The toolkit gave India's delegation a persistent intelligence layer that operated across all active AI governance discussions simultaneously.

Key Findings & Impact

• Deployed 6 geopolitical dashboards covering active AI, digital, and technology governance tracks at the UN
• Automated briefing system reduced morning preparation time by approximately 70%, freeing senior diplomats for strategy
• Negotiation tracker captured amendment-level shifts across G77, EU, and JUSCANZ blocs in near real-time
• India represented in all AI governance discussions at the UN with consistent, data-informed positions
• Tools built for handoff: designed to persist beyond any single advisor's tenure at the Mission

Architecture

Dashboard layer: structured extraction from UN document feeds, delegation statements, and meeting summaries, normalised into a position matrix by country and issue area. Briefing pipeline: overnight batch processing of new inputs, scored by relevance and urgency, rendered into templated briefs for leadership review. Negotiation tracker: amendment-level diff analysis correlated with delegation metadata and historical voting patterns.

Abstract

The rapid expansion of gig platforms raises a central question for labour economists: are these platforms substituting traditional employment or expanding the labour participation frontier? This paper examines the displacement hypothesis across 14 OECD economies using panel data (2012–2023), measuring gig-platform penetration rates against formal employment metrics, worker transitions, and wage distributions. The analysis finds significant heterogeneity: in high-regulation economies (France, Germany), gig growth occurs primarily at the extensive margin, drawing in previously inactive workers. In flexible-regulation economies (UK, US), substitution effects are measurable but concentrated in low-skill service sectors. The paper proposes a regulatory taxonomy distinguishing between additive and substitutive gig growth, with targeted policy recommendations for each regime type.

Key Findings

• Gig-platform penetration explains 8–12% of the variance in part-time employment growth in flexible-regulation economies
• In high-regulation economies, 65% of new gig workers were previously economically inactive, not formally employed
• Wage convergence between gig and formal workers is occurring only in high-skill segments (tech, consulting), not in delivery or logistics
• Regulatory frameworks designed for the industrial era amplify, rather than mitigate, platform labour market distortions

Abstract

Coworking spaces have grown from a niche freelancer phenomenon to a $26B global industry, yet the behavioural drivers of user adoption and retention remain poorly modelled. This study constructs a structural equation model (SEM) decomposing participation into five latent factors: perceived flexibility, community embeddedness, productivity signalling, cost rationalisation, and network access. Using primary survey data from 400+ coworking users across 3 Indian metropolitan areas, the model achieves strong fit (CFI = 0.94, RMSEA = 0.05) and reveals that network access and community embeddedness — not cost savings — are the primary drivers of long-term retention. The findings have direct implications for coworking operators designing retention strategies and for policymakers evaluating shared workspace subsidies as SME infrastructure.

Key Findings

• Network access (loading = 0.82) and community embeddedness (0.76) dominate retention; cost rationalisation (0.41) is a weak predictor
• Productivity signalling mediates the relationship between flexibility and retention, suggesting users derive value from the appearance of structured work
• Users in technology and creative sectors show 2.3x higher community embeddedness scores than traditional professional services
• The model predicts 68% of churn variance, outperforming satisfaction-only models by 22 percentage points

Abstract

The India-UK Free Trade Agreement has been framed as a post-Brexit strategic win for both sides. This paper challenges that framing by conducting a disaggregated cost-benefit analysis at the sectoral level, decomposing projected GDP uplift against strategic concessions on services mobility, data localisation, and intellectual property provisions. Using a computable general equilibrium (CGE) framework calibrated to 2024 bilateral trade flows, the analysis finds that while aggregate GDP gains are positive (0.1–0.3% for India, 0.02–0.07% for the UK), these gains are concentrated in textiles and pharmaceuticals while strategic costs — particularly on Mode 4 services provisions and TRIPS-plus IP commitments — create long-run exposure in India's digital services and generic pharmaceutical sectors, precisely the high-value segments driving India's growth trajectory.

Key Findings

• Aggregate GDP gains mask a distributional asymmetry: 73% of India's projected uplift concentrates in textiles, a declining share of export value
• TRIPS-plus provisions could constrain India's generic pharmaceutical export competitiveness by 4–8% over a 10-year horizon
• Mode 4 services concessions yield less than projected due to UK domestic visa-cap constraints that neutralise treaty commitments
• Strategic cost-benefit ratio inverts when measured against India's revealed comparative advantage trajectory rather than static trade structure

Abstract

India's Pradhan Mantri Fasal Bima Yojana (PMFBY), the world's largest crop insurance programme by enrolment, insures over 120 million farmers yet suffers from a persistent claim-settlement gap: only 35–40% of valid claims are settled within the mandated 45-day window. This paper diagnoses the structural failures driving the gap — adverse selection in voluntary adoption, moral hazard in loss assessment, and institutional bottlenecks at the state-insurer interface — and proposes a redesign framework centred on three interventions: satellite-indexed parametric triggers replacing manual crop-cutting experiments, risk-differentiated premium structures to reduce adverse selection, and performance-linked insurer contracts with automatic penalty escalation. The framework is calibrated using PMFBY administrative data from 6 states (2016–2023), demonstrating that parametric triggers alone could reduce settlement delays by 60% and cut administrative costs by 25%.

Key Findings

• Claim-settlement delays average 127 days (vs. 45-day mandate), with 3 states exceeding 200 days due to manual crop-cutting bottlenecks
• Adverse selection reduces the risk pool quality by 18–22% under voluntary enrolment relative to the pre-2020 mandatory regime
• Satellite-indexed parametric triggers (NDVI-based) achieve 89% correlation with actual yield losses at the gram panchayat level
• Proposed redesign reduces expected fiscal cost per insured farmer by 15% while improving coverage reliability

Abstract

India's Industrial Disputes Act (IDA) of 1947, originally designed to protect worker welfare during post-independence industrialisation, has become the binding constraint on formal manufacturing scale. Chapter V-B, which requires government permission for layoffs in firms with 100+ workers, creates a de facto size ceiling that distorts firm growth incentives. This paper uses a cross-state panel (2001–2022) exploiting variation in state-level IDA amendments to estimate the causal effect of labour regulation rigidity on formal-sector firm size, capital intensity, and labour productivity. Instrumental variable estimation (using colonial-era factory density as an instrument for regulatory stringency) finds that a one-standard-deviation increase in IDA rigidity reduces median formal-sector firm size by 23% and increases capital-labour substitution by 15%, driving India's paradox of capital-intensive manufacturing in a labour-abundant economy.

Key Findings

• Chapter V-B threshold creates a measurable bunching of firms at 99 employees: 34% more firms cluster below the threshold than predicted by the size distribution
• States with reformed IDA provisions (Rajasthan, Andhra Pradesh post-2014) show 18% higher formal manufacturing employment growth
• Capital-labour ratio is 15% higher in rigid-regulation states, suggesting firms substitute machinery for workers to avoid the size threshold
• The informality trap is self-reinforcing: firms that remain small to avoid regulation also forgo scale economies, reducing productivity and tax base

Abstract

The Mahatma Gandhi National Rural Employment Guarantee Act (NREGA), the world's largest public works programme guaranteeing 100 days of wage employment to every rural household, is typically evaluated on income and employment metrics alone. This paper extends the evaluation framework to measure NREGA's second-order effects on democratic participation, institutional trust, and inequality reduction. Using a 15-year panel (2006–2021) covering 450+ districts, the analysis exploits the staggered rollout of NREGA across India's poorest districts (Phase I, 2006) to more developed districts (Phase III, 2008) as a natural experiment. Difference-in-differences estimation shows that NREGA exposure increases gram panchayat electoral participation by 8–12%, women's labour force participation by 6%, and reduces consumption inequality (Gini) by 0.02–0.04 points in treated districts relative to controls.

Key Findings

• NREGA's democratic dividend: 8–12% increase in local electoral participation in Phase I districts, with effects persisting 10+ years after initial rollout
• Women's labour force participation rose 6% in treated districts, driven by guaranteed equal wages acting as a participation floor
• Consumption Gini reduced by 0.02–0.04 points, with the effect concentrated in the bottom two quintiles
• Institutional trust (measured by willingness to engage with local government) increased 14% in high-implementation districts

Abstract

This paper presents two interlocking analyses. First, an econometric decomposition of India's state-level per-capita GDP growth, constructing a growth model that identifies the relative contributions of infrastructure investment, human capital accumulation, institutional quality, and trade openness to cross-state growth divergence. OLS regressions on a 20-year state panel (2001–2021) find that infrastructure quality explains 28% of the cross-state growth variance, more than human capital (19%) or institutional quality (16%). Second, the paper examines the emerging challenge of Chinese steel circumvention through third-country routing, documenting how anti-dumping duties imposed on Chinese steel have redirected export flows through Vietnam, Malaysia, and Indonesia. Trade data analysis reveals that steel exports from these intermediary countries to India surged 140–260% in the 2–3 years following anti-dumping action against China, with product-level analysis confirming that the redirected flows concentrate in the same HS-6 categories targeted by duties.

Key Findings

• Infrastructure quality (roads, power, ports) explains 28% of cross-state per-capita GDP growth variance — the single largest determinant
• Steel exports from Vietnam to India rose 260% within 3 years of anti-dumping duties on Chinese steel, concentrated in duty-targeted HS-6 categories
• Third-country routing accounts for an estimated 40–55% of the circumvention risk in India's flat steel imports
• State-level growth decomposition reveals that Southern and Western states leverage infrastructure quality 2x more effectively than Northern states