Introduction

The Marshall Islands has run one of the more unusual tax systems in the Pacific for decades: no income tax on companies at all, but a Business Gross Receipts Tax charging $80 flat on the first $10,000 of gross revenue and 3% on everything above it, a wages tax that historically started at 8% and rose to 12%, and a personal exemption low enough that ordinary wage earners, not businesses, carried most of the burden. IMF-supported reform work dating to at least 2024 quantified that imbalance precisely: households contributed 67% of total tax collections against 11% from businesses, a ratio the reform agenda aims to move toward 53/23. The Net Profit Tax Act, 2025 (Public Law 38-2025) is the business-side half of that agenda, introducing a genuine profits basis for larger businesses while retaining the BGRT for smaller ones. A March 2026 wage-tax amendment (Bill 103) is the household-side half, raising the tax-free threshold to $8,320 at a cost of roughly $3.1 million in forgone revenue. A broader Marshall Islands Consumption Tax, proposed at 10% to replace scattered import duties and local sales taxes, remains at the policy-note stage.

This reform is happening inside a wider fiscal picture the IMF's December 2025 Article IV consultation described in some detail. The newly renewed Compact of Free Association with the United States, and Trust Fund Committee approval in August 2025 for annual drawdowns from the Compact Trust Fund, now finance a Universal Basic Income and an Extraordinary Needs Distribution targeted at outer-island residents. The IMF's own recommendation is not to stop these transfers but to replace the universal one with something better targeted, and separately, in language unusually direct for an Article IV statement, to abandon a planned digital sovereign bond given the country's acknowledged lack of capacity to manage the risks such an instrument would carry — a caution that echoes RMI's earlier, internationally contested 2018 legislation to create a national digital currency, and that this paper treats as a live governance signal rather than a closed chapter. Fiscal projections swing sharply on the back of these transfers: a 2.4% of GDP surplus in FY2025 gives way to a projected 5.3% deficit in FY2026 as UBI and END roll out, before narrowing again in FY2027 — a path that only makes sense once the CTF drawdown mechanics behind it are made explicit.

The research question this paper takes up has three parts: how the new Net Profit Tax boundary should be monitored so the reform's own revenue-neutrality holds in practice, not only on paper; how much of the CTF's drawdown capacity a universal transfer consumes relative to a targeted one, and what that implies for the fund's multi-decade sustainability; and how RMI's fisheries licence revenue, structurally more reliable than either tax base, should be protected as vessel activity and catch reporting scale up. As throughout this series, the organising identity for the tax-side question splits the gap between potential and actual Net Profit Tax and BGRT revenue into a policy component and a compliance component:

R^{pot}_{biz,t} - R_{biz,t} = \underbrace{\left(R^{pot}_{biz,t} - R^{law}_{biz,t}\right)}_{\text{policy gap: BGRT/NPT threshold design}} + \underbrace{\left(R^{law}_{biz,t} - R_{biz,t}\right)}_{\text{compliance gap: deduction and classification manipulation}}

Current tax issues

2.1 A new boundary between two business tax regimes

The Net Profit Tax Act, 2025 does not replace the BGRT outright; it introduces a net-profit basis alongside it, with smaller businesses expected to remain on the simpler gross-receipts regime. Wherever two regimes sit side by side, the boundary between them becomes something to manage rather than a formality — a business near the line has a reason to structure itself, or its reported turnover, to land on whichever side is more favourable, and a profits basis for the first time creates a deductions question (what counts as a legitimate business expense) that a flat gross-receipts tax never had to ask.

2.2 A tax burden reform is trying to correct, not yet has

The 67/11 household/business split documented in IMF-supported technical work is the reform's starting point, not its current state. The Net Profit Tax and the March 2026 wage-tax threshold increase both move in the intended direction, but neither has been running long enough for the ratio to have shifted materially, and there is, as yet, no published mechanism for tracking whether it does.

2.3 A trust fund now permitted to fund universal transfers

The Compact Trust Fund exists to replace declining Compact grant income over the long run, a design going back to the original 2004 Compact and its own 2009-vintage concern about a 2024 funding cliff. The 2025 Compact renewal and August 2025 Trust Fund Committee decision now permit annual CTF drawdowns to fund a Universal Basic Income and an Extraordinary Needs Distribution. The fiscal path the IMF projects — a 2.4% of GDP surplus in FY2025 swinging to a 5.3% deficit in FY2026 — is a direct mechanical consequence of that decision, and the IMF's own recommendation to replace the universal scheme with a targeted one is, in effect, a request for a drawdown-sustainability calculation the public record does not yet show worked through.

2.4 A digital-asset risk the IMF flagged in unusually direct terms

RMI's 2025 Article IV concluding language recommended the authorities not proceed with a planned global digital sovereign bond launch, citing a lack of the prerequisite capacity and ability to mitigate associated risks. This is not a hypothetical concern for RMI specifically: the country's 2018 legislation to create a national digital currency drew sustained IMF and international criticism at the time over anti-money-laundering and financial-integrity risk, and a fresh digital-instrument proposal revives the same underlying capacity question. For a revenue-intelligence paper, the relevant angle is narrower than monetary policy: any sovereign digital-financial instrument RMI issues needs the same AML/CFT and revenue-integrity safeguards this series applies to conventional tax administration, before launch rather than after.

2.5 Fisheries revenue, structurally more reliable than either tax base

Fisheries, chiefly tuna purse-seine licensing under the regional Vessel Day Scheme (VDS) that RMI participates in as a Party to the Nauru Agreement member, has repeatedly been named by the IMF as a primary driver of growth, and licence-fee revenue is less exposed to the base-erosion and compliance risk that affects the wage and business tax bases. That reliability is itself a reason to protect it: under-reporting of fishing days or catch against licensed allocations is a documented risk across Pacific VDS administrations generally, and RMI's growing dependence on this revenue source raises the cost of any leakage in it.

2.6 Climate exposure and a narrow, emigrating tax base

RMI is a low-lying atoll nation with repeated exposure to droughts and floods, and continued labour outmigration narrows the wage-tax base year on year even as reform tries to make that base fairer. Neither issue is this paper's central focus, but both bound how much any of the three models below can realistically be expected to deliver.

Models

3.1 Model A: Net Profit Tax / BGRT threshold and deduction integrity

Model A has two parts. The first flags businesses sitting near the large/small boundary that separates Net Profit Tax filers from BGRT filers, using the same threshold-proximity logic as the Ghana and Uganda papers in the African series this project's methodology draws on, adapted to a regime split rather than a registration cliff:

\text{BoundaryProximity}_i = 1 - \frac{|\text{GrossRevenue}_i - \Theta|}{\Theta}, \quad \text{for GrossRevenue}_i \text{ near threshold } \Theta

with a network check for turnover split across related entities each kept below Theta, exactly as in the equivalent threshold models elsewhere in this series. The second part addresses the genuinely new risk a profits basis introduces: deductions. A net-profit filer's effective tax rate, compared against its BGRT-equivalent liability had it stayed on gross receipts, should move within a plausible band given typical margins for its sector; a filer whose net-profit liability is implausibly low relative to its gross revenue and sector-typical margin is claiming deductions that warrant review:

\text{DeductionRisk}_i = \max\!\left(0,\; \text{SectorTypicalMargin}_g - \frac{\text{TaxableProfit}_i}{\text{GrossRevenue}_i}\right)

3.2 Model B: Compact Trust Fund drawdown sustainability and UBI-to-targeted comparison

Model B treats the CTF as a depleting-asset problem structurally similar to Timor-Leste's Petroleum Fund in this series' comparative framework, but sized to RMI's much smaller fund and now-permitted annual drawdowns. The fund evolves as:

CTF_{t+1} = CTF_t (1+r) - D_t, \qquad D_t = D^{base}_t + D^{UBI}_t + D^{END}_t

where D-base is the drawdown the fund's design already assumed (replacing declining Compact grants), and D-UBI and D-END are the two new transfer programmes. The comparison the IMF's recommendation implies, but does not itself quantify, is the sustainable drawdown path under each transfer design:

D^{UBI}_t = u \cdot N_{total}, \qquad D^{targeted}_t = u \cdot N_{eligible}, \qquad N_{eligible} \ll N_{total}

where u is the per-person transfer amount and N is population; replacing universal coverage with an eligibility-tested population materially reduces D_t for the same per-person benefit to those who need it most, directly extending the fund's sustainable horizon. The sustainability test itself follows the standard perpetuity-style rule adapted for a finite Compact-support horizon:

\text{Sustainable if } D_t \leq r \cdot CTF_t \text{ once Compact grant support has fully phased out}

3.3 Model C: Vessel Day Scheme revenue-integrity model

Model C reconciles licensed fishing days against reported activity for each vessel operating under RMI's VDS allocation, using satellite Vessel Monitoring System (VMS) position data already required of PNA-licensed vessels:

\text{DaysReported}_v = \big|\{t : \text{vessel } v \text{ recorded fishing activity in RMI EEZ on day } t\}\big|

\text{DaysConsistentWithVMS}_v = \big|\{t \in \text{DaysReported}_v : \text{VMS track confirms presence and fishing-pattern behaviour}\}\big|

\text{IntegrityGap}_v = \text{DaysReported}_v - \text{DaysConsistentWithVMS}_v

A persistent positive gap for a given vessel or operator — self-reported fishing days the VMS track does not corroborate — is a direct revenue-at-risk signal, since VDS fees are charged per licensed day used:

Y_v = \text{DailyVesselFee} \times \text{IntegrityGap}_v

This model does not require RMI to build new satellite tracking; PNA-wide VMS data already exists for exactly this purpose, and Model C's contribution is the systematic day-level reconciliation against it rather than new data collection.

Algorithms and illustrative results

4.1 Algorithm 1: Net Profit Tax / BGRT boundary and deduction screen

ANNUALLY, for each registered business i:
  gross_i     <- declared gross revenue
  regime_i    <- NPT if gross_i >= Theta (or voluntary election), else BGRT
  proximity_i <- 1 - abs(gross_i - Theta) / Theta
  related     <- entities sharing directors/address with i, also near Theta
  IF proximity_i > threshold AND related is non-empty:
     FLAG for entity-resolution review (possible turnover splitting)
  IF regime_i = NPT:
     margin_i       <- taxable_profit_i / gross_i
     sector_margin   <- typical margin for i's sector (from sector aggregate)
     deduction_risk  <- max(0, sector_margin - margin_i)
     IF deduction_risk > threshold: FLAG for expense-substantiation review
PUBLISH aggregate household/business tax-share ratio each year to track
  progress toward the reform's 53/23 target

4.2 Algorithm 2: Vessel Day Scheme reconciliation

FOR each licensed vessel v, fishing season:
  reported_days <- self-reported fishing days in RMI EEZ from vessel logbook
  vms_track     <- satellite VMS position data for v over the season
  vms_days      <- days where vms_track shows presence in RMI EEZ AND a
                    fishing-pattern signature (speed/course consistent with setting or
                    searching, not transiting)
  integrity_gap <- reported_days - vms_days
  IF integrity_gap > tolerance:
     FLAG v for licence-fee reconciliation and, where material, referral to the
     PNA Vessel Day Scheme compliance process
  fee_at_risk   <- daily_vessel_fee * integrity_gap
QUARTERLY: rank flagged vessels/operators by fee_at_risk for follow-up

4.3 Illustrative results: CTF drawdown under universal versus targeted transfers

Using illustrative, clearly hypothetical inputs (a fund sized in the low nine figures, a 5% expected real return, a base drawdown already assumed in the fund's design, an annual per-person transfer, and RMI's approximate population), the model shows the fund's own arithmetic rather than a forecast of the actual CTF, whose real parameters only the Trust Fund Committee holds:

The ratio between the two, not the absolute figures, is the result worth carrying forward: moving from universal to targeted coverage, for the same per-person benefit, roughly doubles-to-triples the sustainable drawdown horizon in this stylised run. That is the trade-off the IMF's recommendation to move toward targeting implies, made explicit and quantifiable rather than left as a qualitative preference.

4.4 Illustrative arithmetic: the tax-burden reform target

If the Net Profit Tax and wage-tax reforms together shift RMI's collection mix from the documented 67/11 household/business split toward the reform's stated 53/23 target over, illustratively, a five-year horizon, and if total tax collections stay roughly flat in real terms (consistent with the reforms' intended revenue neutrality), the arithmetic implies businesses' nominal contribution roughly doubling as a share of total collections while households' falls by about a fifth in relative terms — the rebalancing Algorithm 1's published ratio (above) is designed to track year by year rather than assume has happened once the legislation passes.

Solutions, software, data science and AI models

5.1 Issue-to-solution map

5.2 Model catalogue

5.3 Reference architecture

flowchart TD
  A[Sources: NPT/BGRT returns, CTF and Trust Fund Committee records, PNA VMS feed, VDS licence and catch logs] --> B[Ingestion and data quality]
  B --> C[Warehouse: bitemporal, entity linked]
  C --> D[Boundary and deduction screening service]
  D --> E[Tax case-selection queue]
  C --> F[CTF policy-lab simulator]
  F --> G[Trust Fund Committee dashboard]
  C --> H[VDS reconciliation service]
  H --> I[Fisheries compliance queue]

The VDS reconciliation and NPT/BGRT screening services run on their respective reporting cycles (seasonal for fisheries, annual for tax); the CTF simulator runs on demand for Trust Fund Committee and budget planning.

5.4 Software stack

5.5 Governance of AI use

• No automated reclassification. Boundary and deduction flags route to human review; only that review supports a change in a business's tax treatment.
• CTF simulation is a policy input, not a decision. The Trust Fund Committee, not the model, sets drawdown policy; the model's role is to make the trade-off visible.
• VDS flags go through the existing PNA compliance process. This paper's model feeds that process; it does not replace the regional governance PNA members have already agreed.
• Digital-instrument readiness gate is binary and public. Given the IMF's explicit capacity warning, any future digital sovereign bond proposal should be paired with a published readiness assessment against named risk categories before, not after, a launch decision.

5.6 Team and indicative effort

Given RMI's scale, a very small function — one analyst covering tax data, working with existing Division of Revenue and Taxation staff, plus access to the PNA's regional VMS data-sharing arrangement for Model C — is proportionate. The CTF model (Model B) is a Ministry of Finance and Trust Fund Committee policy tool more than an ongoing operational system, and could reasonably be built as a one-off analytical exercise refreshed annually rather than a standing service. Cost and staffing figures are estimates.

Limitations

• The 67/11 and 53/23 figures are drawn from IMF-supported reform technical work rather than this paper's own analysis; the collection-mix tracking in Algorithm 1 is designed to measure the reform's actual progress against that stated target, not to assume it is already achieved.
• The CTF simulation in section 4.3 uses illustrative, deliberately round-numbered inputs (fund size, return, population, transfer amount); RMI's actual Compact Trust Fund parameters, contribution schedule and Trust Fund Committee rules are more complex than this paper's stylised model and should replace these inputs before any real drawdown decision relies on it.
• Sector-typical margins for the deduction-risk check in Model A need estimation from RMI's own filing data once a full net-profit filing season has occurred; with only one or two years of NPT data available, early estimates will be noisy.
• The VDS reconciliation model assumes access to PNA-wide VMS data under existing regional data-sharing arrangements; the specific access terms RMI has, or would need to negotiate, are outside this paper's knowledge.
• This paper does not take a position on whether a future digital sovereign bond should proceed; it responds only to the IMF's own stated concern that any such instrument should not proceed without the prerequisite capacity, and proposes a governance gate rather than a policy recommendation for or against the instrument itself.

References

• International Monetary Fund. IMF Executive Board Concludes 2025 Article IV Consultation with Republic of the Marshall Islands. Press Release 25/406, 4 December 2025.
• International Monetary Fund. Republic of the Marshall Islands: Staff Concluding Statement of the 2025 Article IV Mission. 10 September 2025.
• International Monetary Fund, Fiscal Affairs Department. Republic of the Marshall Islands: Consumption and Income Tax Reform. Technical Assistance Report, January 2025.
• Nitijela of the Republic of the Marshall Islands. Net Profit Tax Act, 2025 (Public Law 38-2025).
• Pacific Islands Tax Innovation and Technical Assistance (PITI-VITI). Tax Reform Options for the RMI and Estimation of Fiscal and Economic Impact; Adopting a Two-Phased Approach to Tax Reform in the RMI.
• Parties to the Nauru Agreement (PNA). Vessel Day Scheme documentation.
• Methods: as elsewhere in this series — Keen, M. (2013), The Anatomy of the VAT, IMF Working Paper 13/111; Hutton, E. (2017), The Revenue Administration Gap Analysis Program, IMF Technical Notes and Manuals 17/04. Method citations are from memory; verify before publication.