11 Where this paper goes next
This chapter sketches what the paper will look like in its final working-paper form, what is still in progress, and what natural follow-on questions the methodology raises.
11.1 In the working-paper draft (forthcoming on SSRN)
Three tasks remain before the formal preprint:
1. The LIC-eligibility regression discontinuity. Adding tract-level poverty rate and median family income from the American Community Survey enables the canonical NMTC RDD design (Freedman 2012; Harger and Ross 2016) — extended to estimate the rural-versus-urban differential local average treatment effect as a clean causal complement to the descriptive decomposition reported here. Estimated time to add: ~2–3 weeks of data work.
2. Robustness battery. Already partially complete. Includes the unwinsorized leverage version, the top-50-CDE subsample, the pre-2010 versus post-2010 sub-period split, two-way clustering on (CDE, tract), and a USDA-RUCA-based continuous rurality measure replacing the binary metro flag. Reported in §5.3 of the working paper.
3. CDE institutional-form classification. Hand-classifying the top 50 CDEs as bank subsidiary, nonprofit CDFI, for-profit specialized, or governmental — building on the typology in Theodos et al. (2021) — and interacting form with rural status. Lets us pin down the mechanism behind the between-CDE component.
11.2 Natural follow-on questions
Does the within-CDE-equals-zero result generalize to other U.S. place-based subsidy programs? The two natural counterfactuals are LIHTC (Diamond and McQuade (2019); same eligibility logic, different intermediary architecture, longer history) and Opportunity Zones (post-2017, no certified intermediary at all). A multi-program tract-level comparison of mobilization ratios — building on Corinth et al. (2024) — would establish the external validity of the framework within the U.S.
Does it generalize internationally? The methodological framework — leverage as outcome, intermediary FE for decomposition, rural-versus- urban heterogeneity — transfers directly to settings where project- level public-credit-subsidized investments are observable alongside total project cost. Candidates include EU Cohesion Fund deployments (Medeiros 2014; Crescenzi and Giua 2020), EIB project finance (European Investment Bank 2024), and multilateral development bank lending more broadly. Whether the answer is the same is an open empirical question.
What is the optimal NMTC allocation rule? If the binding constraint on aggregate rural mobilization is intermediary selection rather than market structure, the policy lever most likely to move outcomes is the allocation rule — which CDEs receive credits, how much, and on what performance criteria. A formal mechanism-design analysis of the allocation problem, conditional on the empirical decomposition, is a natural next paper.
Does the 20% mandate bind at the allocation-award stage? The current paper finds no bunching at the realized-deployment level. CDFI Fund allocation-award data, were it merged in, would let us test the mandate’s binding behavior at the margin where it is actually enforced.
11.3 Reading the working paper when it ships
The full working paper (forthcoming on SSRN) will have the standard empirical-economics structure:
- Introduction
- Institutional background (Chapter 1 here)
- Data (Chapter 2)
- Empirical strategy (Chapter 5)
- Results (Chapter 6) including the RDD piece
- Discussion (Chapters 7–8)
- Conclusion
The site will be updated with each paper revision; the GitHub repository at https://github.com/ihelfrich/us-nmtc-viewer tracks all changes.
11.4 How to follow updates
Citation form for the work in its current state:
Helfrich, I. T. (2026). The Rural Mobilization Gap in U.S. Place- Based Tax Credit: Intermediary Selection vs. Market Structure in the New Markets Tax Credit. Working paper. https://nmtc.ianhelfrich.com
For substantive comments, corrections, or research collaboration inquiries: contact via the email on the title page or by opening an issue on the GitHub repository.