2  What is the New Markets Tax Credit?

The New Markets Tax Credit (NMTC), enacted under the Community Renewal Tax Relief Act of 2000 and codified at 26 USC §45D, is the U.S. federal government’s largest place-based tax-credit program. It works by offering private investors a 39% federal tax credit, claimed over seven years, in exchange for putting their equity into a certified intermediary called a Community Development Entity (CDE). The CDE then re-deploys that capital as loans or equity into qualifying projects located in low-income census tracts. This chapter walks through the institutional setup with enough precision that the empirical results in later chapters are interpretable.

2.1 The four entities

There are four parties in the NMTC cash flow:

  1. The Investor. A bank, corporation, or wealthy individual with a federal income-tax liability. The investor wants the 39% credit and places their equity into a CDE.

  2. The Community Development Entity (CDE). A regulated intermediary certified by the CDFI Fund, the U.S. Treasury division that administers NMTC. Roughly 600 active CDEs operated over the FY2001– FY2022 sample period, ranging from bank subsidiaries (e.g., USBCDE LLC, Banc of America CDE) to nonprofit community-development financial institutions (e.g., Local Initiatives Support Corporation, Reinvestment Fund) to specialized rural CDEs (e.g., Rural Development Partners LLC, Montana Community Development Corporation). The CDE accepts the investor’s equity — called a Qualified Equity Investment, or QEI — and is required to deploy substantially all of it (at least 85%) into qualifying low-income community businesses within twelve months.

  3. The Qualified Active Low-Income Community Business (QALICB). The ultimate recipient — a real-estate development, an operating business, a special-purpose entity, or another CDE in a pass-through arrangement. The CDE’s deployment to the QALICB is called a Qualified Low-Income Community Investment, or QLICI. This is the level that the public CDFI Fund data records.

  4. The federal government — specifically Treasury, which issues the credit through the IRS, and the Community Development Financial Institutions (CDFI) Fund, which administers the program (allocates credits to CDEs, certifies CDEs, monitors compliance, and publishes the annual transaction data release used in this paper).

2.2 The cash flow

                    QEI $                         QLICI $
   Investor  ────────────────►  CDE  ────────────────────►  QALICB
            ◄──────────────                                  (project /
              39% tax credit                                 business in
            (over 7 years)                                   LIC tract)

What the public data records is the second arrow — the QLICI flow from CDE to QALICB. The first arrow (the QEI from investor to CDE) is governed by securities-law disclosures, not CDFI Fund disclosures, and is not in our analytical sample.

2.3 The eligibility rules

A census tract qualifies as a “Low-Income Community” (LIC) — and is therefore NMTC-eligible — if either of the following holds:

\text{LICeligible}_\ell \;=\; \mathbf{1}\!\left\{ \text{Poverty}_\ell \geq 0.20 \;\;\vee\;\; \frac{\text{MFI}_\ell}{\text{AreaMFI}_\ell} \leq 0.80 \right\}

In words: the tract qualifies if poverty rate ≥ 20%, or median family income ≤ 80% of area median family income. Either condition alone is sufficient. There are also “targeted population” provisions and high-migration-rural-county overrides that we treat as part of the residual variation.

The 20% poverty cutoff is what makes NMTC amenable to a sharp regression-discontinuity design: tracts at 19.9% poverty are nearly identical to tracts at 20.1% poverty in everything except eligibility status. Freedman (2012) is the canonical paper using this design; Harger and Ross (2016) is the second canonical reference.

2.4 The 20% non-metropolitan statutory mandate

The statute requires CDEs to direct at least 20% of their QLICIs to non-metropolitan census tracts (Public Law 106-554 §121, codified at 26 USC §45D). This rule was added because, without it, CDEs were expected to concentrate deployments in urban areas where deal flow is denser and underwriting is easier. The 20% is enforced at the allocation-award stage: a CDE must demonstrate intent to deploy at least 20% rurally as part of its competitive application for an NMTC allocation award.

In our data — which captures deployments (QLICIs), not allocation commitments — non-metro tracts received exactly 19.6% of dollars and 19.3% of transactions over 2001–2022. That’s just under 20%, suggesting the mandate binds in some sense at the program level. Whether it binds at the individual CDE level is something we test directly via a bunching diagnostic.

2.5 QALICB types

The CDFI Fund classifies every project into one of four QALICB types. The classification matters because different types have very different leverage capabilities:

code meaning leverage profile n
RE Real Estate Stackable: mezzanine debt, second mortgages, equity tranches all possible. Highest median leverage in our data. 2,862
NRE Non-Real-Estate (operating business) Less collateral, harder to leverage. 3,775
SPE Special-Purpose Entity Flexible structurally, typically simpler stacks. 1,296
CDE Loan-to-CDE Mechanically pinned at 1.0× by structure. 91

The cross-tabulation by metro status is informative: real-estate is ~40% of metro projects but only ~17% of non-metro; operating businesses (NRE) are ~50% of metro and ~65% of non-metro. Project-type composition itself differs between rural and urban — a confounder our regression strategy explicitly addresses.

2.6 Why this institutional detail matters for the empirical strategy

Three features of the institutional setup are load-bearing for the analysis in Chapter 5:

  1. CDEs are heterogeneous and persistent. A CDE applies for an NMTC allocation award, gets one (or doesn’t), then deploys it over several years. The same CDE often does both metro and non-metro deals. This persistence is what makes within-CDE fixed-effects identification feasible.
  2. The 20% rule creates an institutional reason to expect bunching. If the rule binds at the cumulative-deployment level, we should see CDEs piled up at exactly s_j = 0.20 in the cross-CDE distribution.
  3. The LIC eligibility cutoff creates an exogenous source of program access. Tracts that just barely qualify are otherwise similar to tracts that just barely don’t.