Local areas overseen by multiple elected representatives see more firms, jobs and economic activity, as mutual oversight trims cronyism and regulatory hold-ups; the gains are largest when politicians are partisan rivals, consistent with checks-and-balances constraining rent extraction.

Main Finding

Areas governed by multiple elected representatives causally experience greater economic activity: being in a “split” administrative block (one that is split across multiple electoral constituencies) raises formal firm entry, employment, and local nightlight intensity. The effect operates through strengthened checks-and-balances that reduce rent extraction, regulatory hold-up, and cronyism in implementation.

Key Points

• Identification strategy
  • Exploits haphazard overlap between administrative Blocks and electoral Assembly Constituencies in India: some Blocks are entirely inside a constituency (unsplit) while others are split across multiple constituencies (split).
  • Primary design: geographic regression discontinuity (RD) comparing villages just inside split blocks to those just outside in adjacent unsplit blocks.
  • Secondary design: differences-in-discontinuity leveraging the 2008 nationwide delimitation (which changed constituency boundaries), comparing blocks that switch between split/unsplit status and suitable controls.
• Magnitudes & core outcomes
  • Cross-sectional RD: split blocks → ≈3% higher firm entry and ≈7% higher nightlight intensity.
  • Differences-in-discontinuity (post-delimitation): transitions from single to multiple representatives → 1.1–1.6% increase in firm entry; reverse transitions → 0.7–1.0% decrease.
  • Labor outcomes: villages in split blocks show ≈5% higher employment and lower application rates to NREGA (interpreted as better private labor market options).
• Mechanisms and heterogeneity
  • Mutual oversight: multiple MLAs jointly observe the same Block Development Officer (BDO) in formal meetings, increasing monitoring and making exclusive rent-sharing hard.
  • Partisan non-alignment amplifies effects: split blocks governed by politicians from different parties see larger increases in firm entry (consistent with stronger incentives to expose rivals).
  • Politicians facing more mutual oversight accumulate less private wealth: constituencies with a higher share of split blocks show a lower winner’s premium (post-election asset growth of winners vs runners-up).
  • Firm-entry margins consistent with reduced unchecked discretion:
    • Preferential treatment/cronyism declines: connected (co-caste) firms and politically-favored industries show relatively lower entry in split blocks.
    • Regulatory hold-up eases: projects in split blocks obtain approvals faster; regulated industries and small firms—more sensitive to discretionary delays—experience larger relative gains.
    • Public goods provision improves: PMGSY road cost overruns fall when blocks become split; villages in split blocks get more reliable commercial electricity.
• Robustness / threats addressed
  • Balance checks using 2001 Census covariates show smoothness across boundaries.
  • Restricting to straight-line/natural-feature boundaries yields similar estimates.
  • 2008 Delimitation plausibly exogenous; prior literature suggests limited gerrymandering in that process.

Data & Methods

• Data sources
  • Firm entry: Ministry of Corporate Affairs (MCA) corporate registry (precise registration addresses, geocoded).
  • Nightlights: satellite-based nighttime lights as local activity proxy.
  • Employment & program demand: employment measures and NREGA application records at local levels.
  • Public works: PMGSY road project records (costs, overruns).
  • Political data: MLA identity, party affiliation, and candidate sworn asset affidavits (used to compute winner’s premium).
  • Patronage proxies: surname→caste (jati) mapping from People of India Project used to identify possible co-ethnic/connected firms.
• Empirical designs
  • Geographic RD: comparing observations very close to split–unsplit block boundaries, assuming smoothness of confounders.
  • Differences-in-discontinuity: using the timing of 2008 delimitation to observe within-location changes as blocks switch split status.
  • Heterogeneity and mechanism tests: interaction analyses by partisan alignment, firm size, industry regulatory intensity, and constituency share of split blocks (for winner’s premium).
• Key identification assumptions and tests
  • Smoothness of potential confounders at block boundaries — supported by balance tests.
  • No selective sorting by delimitation — supported by prior work and robustness subsamples.
  • Dose-response (more representatives → larger effects) and symmetry of switches (unsplit→split vs split→unsplit) bolster causal interpretation.

Implications for AI Economics

• Institutional design matters for market entry and innovation: The paper shows that multi-principal political oversight reduces discretionary capture, lowering entry barriers. AI economics models that predict firm entry, competition, and innovation should incorporate political-institutional constraints (single vs multiple principals) as determinants of incumbents’ ability to create entry barriers or extract rents.
• Regulatory capture and approval processes for AI: The finding that mutual political oversight speeds regulatory approvals and reduces hold-up implies that governance structures for AI (standalone agencies vs multi-stakeholder oversight) will materially affect time-to-market, vendor concentration, and the degree of favoritism in public procurement of AI systems. Multi-principal or cross-party oversight mechanisms could reduce capture and encourage diverse AI firm entry.
• Heterogeneous effects by firm type & sector: Smaller firms and highly regulated industries benefit more from checks-and-balances. For AI, this suggests that multi-principal governance could particularly help small AI startups and regulated AI applications (healthcare, finance) by reducing discretionary delays and cronyism that favor large incumbents.
• Politicians’ incentives shape tech policy & diffusion: The winner’s-premium evidence links politicians’ private returns to the degree of oversight. Modeling political economy of AI regulation should account for how incumbent gains from capture affect policy stringency, procurement choices, and enforcement—kinds of rents matter for predicted AI adoption trajectories.
• Empirical strategy for AI policy evaluation: The paper demonstrates a strong causal identification strategy using administrative/electoral boundary misalignment. AI economists can adapt similar geographic RD or delimitation-based difference-in-discontinuity designs to study causal effects of political/institutional variation on AI adoption, digital infrastructure investment, local data governance, or outcomes of AI-targeted public programs.
• Policy takeaways for AI governance:
  • Multi-principal monitoring (e.g., inter-agency or bipartisan oversight) can reduce capture in AI procurement and implementation.
  • Designing institutional checks—especially when partisan non-alignment is present—can improve competitive entry, reduce favoritism, and speed approvals for responsible AI systems.
• Research avenues
  • Use analogous boundary/discontinuity designs to measure how political-institutional variation affects AI firm entry, adoption of AI in SMEs, or the distributional labor impacts of AI across localities.
  • Study whether multi-principal oversight affects AI-specific metrics: algorithmic transparency, procurement outcomes, vendor diversity, or regulatory compliance timelines.
  • Examine whether checks-and-balances alter the incidence of algorithmic favoritism or procurement cronyism in AI projects funded or overseen by local governments.

Limitations to bear in mind when translating to AI contexts: external validity beyond India’s formal corporate sector and BDO/MLA institutional details may be limited; informal firms and sectoral differences matter; mechanisms are supported by suggestive evidence but remain complex and context-dependent.